Introduction

The incidence and economic burden of head and neck squamous cell cancers (HNSCC) in India are disproportionately higher than other regions, making the disease a significant public health concern. Over 80% of these patients present with locally advanced disease1. According to GLOBOCAN 2022, HNSCC, spanning lip and oral cavity, oropharynx, hypopharynx, larynx and nasopharynx, accounted for an estimated 892,000 new cases (4.5% of all cancers) and 458,000 deaths (4.7% of all cancer deaths), underscoring its major public-health burden2. Over the current decade, HNSCC is projected to cost the global economy nearly US$ 535 billion in lost output, largely driven by productivity losses and the expense of complex multimodality care3.

The management of borderline resectable or inoperable HNSCC poses a formidable challenge, especially since the cost of treating advanced stages is 42% greater than that in the early stages (US$ 2717 vs. US$ 1568) underscoring the universal need for innovative strategies that optimize immunotherapy efficacy while improving cost efficiency and accessibility across diverse healthcare settings4. HNSCCs, especially unresectable oral cavity squamous cell cancers (OCSCC) are treated with concurrent chemoradiation where feasible or radiation therapy alone. The traditional approach of surgery followed by adjuvant therapy, although effective, is not viable for patients with unresectable disease due to the risk of significant morbidity and the impact on physiological function and quality of life. This treatment approach without surgical resection can be considered palliative, with a 5-year disease-specific survival of 12% which is closer to that of patients with recurrent/metastatic HNSCC5. The randomised trials of induction chemotherapy prior to surgery or chemoradiation have failed to demonstrate an overall survival benefit when compared to standard therapy6. However, an induction(neoadjuvant) approach followed by assessment for radical local therapy (surgical resection or chemoradiation) has been employed in India with promising results7,8.

The advent of immune checkpoint inhibitors (ICIs) has revolutionized the treatment landscape for HNSCC9. In the recurrent/metastatic setting, anti-PD-1 monotherapy after platinum failure improved median overall survival (OS) from 5.1 to 7.5 months with nivolumab, and up to 8 months with pembrolizumab, compared to standard chemotherapy10,11. In the first-line setting, pembrolizumab combined with platinum-fluoropyrimidine chemotherapy extended median OS to 13.0 months (vs. 10.7 months with cetuximab-chemotherapy) and achieved an overall response rate (ORR) of 43% vs. 38% in the CPS ≥ 20 cohort12.

The success of ICIs in this setting has prompted investigation into their potential role in earlier disease stages, including as part of induction regimens prior to definitive chemoradiation or surgical resection13,14,15. Despite the proven efficacy of ICIs, their high cost poses a significant barrier to access, particularly in health systems where patients pay out-of-pocket. It is estimated that less than 5% of patients in these regions with head and neck cancers who need them have access to ICIs at approved doses16.

This disparity underscores the urgent need for innovative strategies that optimize immunotherapy efficacy while improving cost efficiency and accessibility across diverse healthcare settings. One such strategy has been to use nivolumab at lower than the approved dose of 240 mg every 2 weeks for similar efficacy at a fraction of the cost. The initial phase I studies of nivolumab showed that very low doses had clinical activity and led to receptor saturation similar to that observed with higher doses17,18. Thus the rationale for using a lower dose resonates with the principle of the optimal biological dose rather than the traditional maximum tolerated dose(MTD) for such biological agents. Patil et al. have already shown, in a randomised phase 3 study, that very-low-dose nivolumab (20 mg every 3 weeks) plus triple metronomic chemotherapy (TMC-I) improves OS versus standard chemotherapy alone in R/M HNSCC19.

In response to escalating treatment costs and the need for cost-efficacy balance and building upon the success of low dose nivolumab in the recurrent/metastatic setting, we employed a novel regimen incorporating low-dose nivolumab with standard induction chemotherapy for patients who present with locally advanced inoperable or borderline resectable head and neck cancers. Our aim was to maintain efficacy while reducing treatment expense- a strategy with potential benefits in both resource-limited as well as resource-rich settings.

This ambispective study was undertaken to evaluate the potential of low-dose immunotherapy as part of a multimodal treatment strategy, to improve response rates and conversion to local therapy, with the ultimate goal of improving patient outcomes and offering a more sustainable model for the use of expensive immunotherapeutic agents in the treatment of HNSCC.

Patients and methods

Study design and patient selection

We conducted a single-centre ambispective cohort study of patients with HNSCC treated at our institution, a 3675-bed, quaternary referral, university-affiliated academic hospital in South India. Initial clinical data were collected retrospectively from electronic medical records, and survival data were collected prospectively through telephonic follow-up. Patients diagnosed with locally advanced (stage III-IVB) non-metastatic HNSCC and deemed unresectable/borderline resectable at initial presentation following discussion in the multidisciplinary tumour board, were included. Eligibility criteria included patients aged ≥ 18 years with histologically confirmed squamous cell carcinoma, an ECOG performance status of 0–2, and having received at least one dose of low dose nivolumab as part of the induction regimen. The criteria for inoperability used at our institution for HNSCC are extensive involvement of the skin or diffuse peritumoral oedema, high infratemporal fossa involvement, significant involvement of the posterior third of the tongue, skull base invasion, fixation to the spine or prevertebral fascia or muscles, and carotid and/or brachial plexus encasement. Exclusion criteria included prior systemic therapy or radiotherapy for HNSCC, recurrent or metastatic disease, active autoimmune disease requiring immunosuppression, uncontrolled intercurrent illness (grade ≥ 2 heart failure, active hepatitis B/C, HIV with CD4 < 200 cells µL−1), baseline grade ≥ 2 peripheral neuropathy, pregnancy or lactation and any contraindications to chemotherapy or immunotherapy.

Eligible patients were identified from our institution’s prospectively maintained head and neck tumour board database that includes comprehensive records of all HNSCC cases discussed. We retrospectively reviewed and extracted individual patient data from the electronic medical records (EMR) and the institutional comprehensive laboratory information system (LIS). The collected data encompassed baseline demographic information, disease characteristics, treatment details, dates of key events, toxicity data (CTCAE v5.0) and status at last outpatient follow-up.

This study was conducted in accordance with the guidelines established by the International Conference on Harmonisation Good Clinical Practice (ICH-GCP), the Declaration of Helsinki, and the Code of Conduct set forth by the Indian Council of Medical Research (ICMR). The study protocol was approved by the Institutional Review Board (Ethics Committee). As the study was retrospective and involved data collected during routine clinical care and follow-up, the requirement for informed consent was waived by the Institutional Ethics Committee- The Institutional Review Board (IRB) of Christian Medical College (CMC), Vellore [Ref: IRB no. 15358 dated 19.4.2023.].

Sample size estimation

The sample size for our ambispective single-arm cohort study was determined based on the primary endpoint of overall response rate. We employed the one-sample binomial test to evaluate whether the ORR observed with the study regimen was significantly higher than historical controls. In the null hypothesis, we assumed an ORR of 45%, based on previously published outcomes with standard induction chemotherapy in comparable patient populations. This figure represents the pooled ORR observed with platinum-taxane doublet and triplet induction regimens across multiple Indian series7,8,20,21.

In the alternative hypothesis we selected an ORR of 60%. This absolute improvement was deemed clinically meaningful because it would translate to 15 additional responders per 100 treated, reaching an ORR (for this unresectable cohort), comparable to that seen with resectable tumours in the TAX 323, TAX 324 and subsequent induction trials22,23.

Using the one-arm binomial method, at α of 0.05 and a power of 90%, the minimum required sample size was calculated to be 95 patients (CRAB SWOG Statistical Tools Calculators (https://stattools.crab.org)). A total of 111 patients met the eligibility criteria and were included in the final analysis, slightly exceeding the calculated sample size. This ensured sufficient precision for estimation of the primary endpoint and allowed for exploratory subgroup analysis.

Treatment protocol

Patients received induction chemotherapy combined with low-dose nivolumab. The primary chemotherapy regimen was a combination of taxane-platinum +/− fluoropyrimidine (TPF) in the majority of patients. This included carboplatin-paclitaxel, docetaxel-cisplatin-5-FU, docetaxel-cisplatin-capecitabine, or docetaxel-cisplatin. For patients deemed unfit for these standard regimens, a triple oral metronomic chemotherapy was administered, comprising weekly methotrexate (15 mg m−2), celecoxib (200 mg twice daily), and erlotinib (150 mg once daily). Nivolumab was administered at low doses, defined as < 240 mg or < 3 mg kg−1 every 2 weeks. Specific dosing schedules varied, including 20 mg, 40 mg, or 100 mg given every 2 to 4 weeks with individual patient doses selected based on logistical and financial feasibility. When very low doses were administered, vials containing the lowest possible doses were opened, refrigerated, and reused for subsequent cycles. We developed this protocol based on the evidence of stability of nivolumab in its original vials after opening, for 4 weeks at 2–8 °C24.

Outcome measures

The main efficacy outcome we studied was overall response rate (ORR). The other outcomes assessed were progression-free survival (PFS), overall survival (OS), toxicity profile, and cost analysis. PFS was defined as the time from diagnosis to disease progression or death from any cause, while OS was calculated from diagnosis to death from any cause. We employed a two-tiered approach that included review of EMR for documented clinical visits and prospective telephonic interviews with patients or their caregivers for those without recent clinical visits. For patients who underwent surgical resection, the pathological complete response(pCR) was assessed. Toxicity was evaluated using the National Cancer Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.

Response assessment

Tumour response was evaluated using CT and/or MRI according to RECIST v1.1 criteria. Two independent radiologists assessed the response, and any discrepancies were resolved by a consensus meeting. In the meeting, both radiologists reviewed the scans together, discussed the differences in their individual assessments, and reached an agreement on the final response classification. If a consensus could not be reached, a third senior radiologist was consulted to provide the final decision.

Cost analysis was conducted to assess the direct costs of utilizing a low-dose nivolumab regimen compared to standard dosing. The median absolute dose of nivolumab administered per patient was calculated based on the dosing frequency and total number of doses received. The median cost per patient was derived by multiplying the median dose by the cost per milligram, which was determined based on the current pricing at our centre for available vial sizes (40 mg and 100 mg). The percentage reduction in drug cost per patient when using the low-dose approach versus standard regimens was also calculated.

Statistical analysis

Descriptive statistics were used to summarize patient demographics and clinical characteristics. Categorical variables were compared using Chi-square or Fisher’s exact tests, while continuous variables were compared using independent t-tests or Mann–Whitney U tests. Normality of continuous variables was first assessed visually with histograms and Q–Q (quantile–quantile) plots, and then formally tested using the Shapiro–Wilk test; non-normal data were summarised as medians with interquartile ranges (IQRs) and compared with the Mann–Whitney U test.

Survival outcomes, including progression-free survival (PFS) and overall survival (OS), were estimated using the Kaplan–Meier method, with comparisons between groups performed using the log-rank test. A p-value of < 0.05 was considered statistically significant. All statistical analyses were conducted using IBM SPSS Statistics v26 and R v4.1.2. Graphs and figures were generated using the ggplot2 and survminer packages in R v4.1.2.

Results

Patient characteristics

A total of 111 patients with locally advanced inoperable non-metastatic HNSCC diagnosed between Sep 1, 2020, and May 26, 2023, were included in this study. The cohort primarily consisted of male individuals (78%), with a median age of 48 years (Table 1). The predominant tumour site was oral cavity (79%) followed by hypopharynx, oropharynx and larynx. The majority of patients presented with stage IVA or IVB disease (97%). 64% of the patients reported a history of smoking at some point; however, data was unavailable for nearly 12% of the cases.

Table 1 Baseline demographic, clinical and treatment characteristics of the study cohort (N = 111).
Full size table

Treatment delivery

The predominant induction chemotherapy regimens were platinum doublets, administered to 69% of patients, and triplet regimens, given to 26.1% of patients. A minority of patients who were unfit for the above regimens received triple OMCT as detailed earlier. The dosing of nivolumab varied based on multiple factors, primarily financial feasibility. To maintain consistency in comparison, we transformed the nivolumab dosages to a weight-based biweekly basis. After conversion, the harmonized median dose of nivolumab was 0.51 mg kg−1 every 2 weeks. The median number of doses administered was 3 (IQR 2–4).

Efficacy outcomes

Radiologic tumour response

RECIST v1.1 response assessment was possible for 89 out of 111 patients. The remaining patients either did not have good quality scans done at baseline for comparison or were lost to follow up prior to the first reassessment scan. Among the entire cohort 5 patients (5.6%) had complete response (CR), 62 patients (69.7%) had a partial response (PR), 19 patients (21.3%) had stable disease (SD) and 3 patients (3.4%) had progressive disease (PD) as illustrated in Fig. 1. The overall response rate (ORR) among patients who were evaluated was 75.3% (67/89). The overall response rate among the entire population treated, including those without reassessment was 60.4% (67/111). The site-specific ORR varied, with 100% ORR seen in larynx primaries and oral cavity tumours having the lowest ORR at 69.8% (Fig. 2). There was a significant but weak correlation between total dose of nivolumab and change in tumour from baseline by RECIST (p = 0.03, r = − 0.23; Fig. 3). There was no significant correlation between the harmonised median dose of nivolumab and the change in tumour from baseline by RECIST (p = 0.78, r = − 0.03). There was no difference in the harmonised median dose of nivolumab nor the total dose of nivolumab between those who had an objective response vs. those who did not (p = 0.86, p = 0.28 respectively).

Fig. 1
figure 1

Waterfall plot of best percentage change in target-lesion sum from baseline after induction therapy in evaluable patients (n = 89). ORR, overall response rate; RECIST, Response evaluation criteria in solid tumours.

Full size image
Fig. 2
figure 2

Site-specific overall response rate (ORR) for the evaluable cohort (n = 89). The primary tumour site is plotted on the x-axis; ORR (%) is shown on the y-axis.

Full size image
Fig. 3
figure 3

Scatter plot of cumulative nivolumab dose (mg, x-axis) versus best percentage change in target-lesion sum (y-axis). The solid line represents a LOESS fit; shaded area shows the 95% confidence interval.

Full size image

Conversion to radical therapy and pathological response

Of the 79 patients with oral cavity tumours, 25 (31.6%) patients had conversion with their tumour becoming resectable following induction therapy and rediscussion in the tumour board. Among these patients who underwent resection of their tumours, 8 patients (32% of resected; 10% of all oral cavity tumours) achieved pathological complete response (pCR) at the time of surgery.

Among non-oral cavity tumour sites, the following proportions of patients proceeded to radical radiotherapy or chemoradiotherapy after induction: 60% of patients with oropharyngeal tumours, 69% of patients with hypopharyngeal tumours, and 77.8% of patients with laryngeal tumours. Overall, in non-oral cavity tumour sites the conversion rate to radical therapy was 68.8%. The treatment pathway is summarised in Fig. 4.

Fig. 4
figure 4

Sankey diagram of flow of therapy following induction. Node widths are proportional to patient numbers.

Full size image

Survival outcomes

At a median follow up of 16 months in the entire cohort, the median progression-free survival and overall survival was not reached. The 1-year PFS rate was 66.9%, and the 1-year OS rate was 83.1%. Patients who had any radical therapy including surgical resection or radical chemoradiation demonstrated significantly longer PFS as compared to those who did not (NR vs. 12.2 months; p < 0.0001; Fig. 7). The PFS was significantly longer for patients with either a pCR or radiological complete response (rCR), versus those without rCR/pCR (NR vs. 18.6 months; p < 0.01; Fig. 6). No recurrences or deaths were observed in the pCR group at the time of last follow up. Patients receiving nivolumab doses less than 0.6 mg kg−1 had a numerically shorter PFS compared to those receiving doses higher than 0.6 mg kg−1 (17.9 months vs. NR; p = 0.11; Fig. 5). Patients who received a triplet chemotherapy backbone demonstrated a longer PFS as compared to those who received other regimens (NR vs. 18.6 months; p = 0.15). To assess the robustness of our dose-PFS analysis, we performed sensitivity analyses using alternative dose thresholds of 0.4 mg kg−1, 0.5 mg kg−1, and 1.0 mg kg−1. At these thresholds, no significant differences in median PFS were observed between the higher and lower dosing groups (p = 0.86, 0.97, and 0.72, respectively).

Fig. 5
figure 5

Kaplan–Meier progression-free survival (PFS) stratified by nivolumab dose (< 0.6 mg kg−1 vs. ≥ 0.6 mg kg−1). Time (months) from diagnosis is on the x-axis; cumulative probability of remaining progression-free is on the y-axis. Vertical marks denote censored observations.

Full size image
Fig. 6
figure 6

Kaplan–Meier PFS by radiological or pathological complete response (rCR/pCR) status. Time (months) from diagnosis is on the x-axis; cumulative probability of remaining progression-free is on the y-axis. Vertical marks denote censored observations.

Full size image
Fig. 7
figure 7

Kaplan–Meier PFS according to receipt of radical therapy after induction. Time (months) from diagnosis is on the x-axis; cumulative probability of remaining progression-free is on the y-axis. Vertical marks denote censored observations.

Full size image

Safety profile

The spectrum and grade of adverse events are summarised in Table 2, with the most frequent events being haematological. The most frequent adverse event was anaemia, affecting 69 patients (62.2%), predominantly grade 1 or 2. Grade 3 anaemia requiring blood transfusion was seen in 7 patients (6.3%), with no grade 4 anaemia being reported. Leukopenia was observed in 34 patients (30.6%), that was grade 3 in 5 patients (4.5%) and grade 4 in 1 patient (0.9%). Neutropenia was seen in 18 patients (16.2%), being grade 3 in 5 patients (4.5%) and grade 4 in 6 patients (5.4%). Febrile neutropenia was rare, occurring in only 1 patient (0.9%).

Thrombocytopenia was noted in 19 patients (17.1%), which was grade 3 in 3 patients (2.7%). 32 patients (28.8%) had leukocytosis, which was grade 3 in 8 (7.2%) and grade 4 in 2 (1.8%). Hyponatremia was the most common non-haematological adverse event, affecting 36 patients (32.4%), mostly grade 1 (23.4%). Grade 3 hyponatremia was observed in 4 patients (3.6%).

Immune-related adverse events (iRAE) included hypothyroidism that was observed in 13 patients (11.7%), with grade 3 in 2 patients (1.8%) and grade 4 in 2 patients (1.8%). Hyperthyroidism was seen only in 6 patients (5.4%) which was grade 1 or 2. Elevated alanine aminotransferase (ALT) levels were observed in 17 patients (15.3%), with grade 4 elevation in 1 patient (0.9%). Elevated aspartate aminotransferase (AST) levels occurred in 11 patients (9.9%), with grade 3 and 4 elevations in 1 patient each (0.9%). No patient developed colitis, pneumonitis, myocarditis or neurological iRAEs.

Other significant adverse events that were documented included chemotherapy-induced nausea and vomiting in 8 patients (7.2%), oral mucositis in 7 patients (6.3%), diarrhoea in 5 patients (4.5%), fatigue in 2 patients (1.8%), and rash in 1 patient (0.9%). The majority of these events were of grade 1 or 2 severity.

Overall, grade 3 or higher adverse events were reported in 35 patients (31.5%). There were no treatment-related deaths.

Table 2 Treatment-emergent adverse events, including immune-related adverse events (irAEs), graded according to CTCAE v5.0.
Full size table

Cost analysis

The median absolute dose of nivolumab was 26.67 mg per administration. As the median number of doses per patient was 3, the median total nivolumab dose per patient was about 80 mg. Considering the cost of nivolumab at our centre of USD 485.37 for a 40 mg vial and USD 1213.41 for a 100 mg vial, the cost per milligram is approximately USD 12.13. Therefore, the median cost of nivolumab per patient in our study was approximately USD 970.4. The standard dosing regimen for nivolumab is 240 mg administered every two weeks. Over three doses, this amounts to a total of 720 mg per patient, with an associated cost of approximately USD 8733.6. This low-dose regimen reduced the nivolumab drug cost per patient by approximately 88.9%. Employing similar analysis, compared to the pembrolizumab regimen as per KEYNOTE-048, this approach reduced the cost per patient by approximately 93.9%.

Discussion

Our ambispective study in a resource-limited setting, assessed adding low-dose nivolumab to induction chemotherapy in patients with locally advanced, unresectable/borderline resectable non-metastatic head and neck squamous cell carcinoma (HNSCC). We report a promising ORR of 75.3%. Among patients with oral cavity tumours, 31.6% achieved conversion to resectability, with one-third of them attaining pCR at surgical resection. The 1-year PFS of 66.9%, and the OS of 83.1% are encouraging when compared to historical controls of patients with unresectable HNSCC25. Treatment was generally well-tolerated; the most common adverse events were manageable haematological toxicities like anaemia, leukopenia, and neutropenia. Low-grade iRAEs including hepatitis, hypothyroidism and hyperthyroidism, were observed. The observed conversion rate and pCR can significantly improve patient outcomes, offering survival gains especially in oral cavity cancers.

To the best of our knowledge this study is the first report in literature of the efficacy and safety of low dose nivolumab with IC in this patient population. Our results indicate that even at reduced doses, nivolumab may enhance chemotherapy’s anti-tumour activity through immunomodulatory effects, supporting the hypothesis that lower doses of immune checkpoint inhibitors can achieve adequate receptor occupancy for an anti-tumour immune response. Previously, phase 3 trial data has also demonstrated that low-dose nivolumab (20 mg every 3 weeks) added to triple oral metronomic chemotherapy improves OS in patients with recurrent/metastatic HNSCC19.

We found a possible association between nivolumab dose and treatment outcomes with patients receiving doses < 0.6 mg kg−1 having a numerically shorter median PFS. This dose-dependent efficacy suggests a potential threshold below which survival outcomes may be compromised. This raises pertinent questions about the optimal dose and schedule of nivolumab in the neoadjuvant setting, especially considering cost implications in resource-constrained settings.

Our findings support the concept that the efficacy of immune checkpoint inhibitors like nivolumab may follow a receptor occupancy model rather than a traditional dose-response relationship. The saturation of PD-1 receptors on T cells can occur at doses substantially lower than the approved dose, as demonstrated in pharmacodynamic studies showing that doses as low as 0.1–0.3 mg kg−1 achieve > 70% receptor occupancy with minimal additional binding at higher doses17,18. Despite sub-saturation, this correlated with sustained T-cell activation and antitumour responses26,27. Nivolumab also shows prolonged occupancy at ~ 70% even when nivolumab serum concentrations fell below quantification limits (< 1 µg mL−1). The mechanism of action of nivolumab may require only threshold receptor occupancy rather than complete saturation to trigger immune activation. Using it in combination with chemotherapy may further enhance this effect through immunogenic cell death, releasing tumour antigens and damage-associated molecular patterns that can potentiate the immune response even with lower nivolumab doses28. The inter-patient variability in PD-1/PD-L1 expression and the difference in tumour microenvironment characteristics may explain why some patients have robust responses at low doses while others require higher exposure. These biological principles suggest that the prevailing paradigm of “more is better” may not apply to immune checkpoint inhibitors, potentially allowing us to attempt significant dose de-escalation without compromising efficacy.

Our findings align with prior studies evaluating induction chemotherapy in early and locally advanced HNSCC. Noronha et al. demonstrated that neoadjuvant chemotherapy could render borderline resectable oral cavity tumours (T4a and T4b) amenable to resection, with an ORR of 32.5%, and 32% undergoing surgery after NACT8. Our cohort with unresectable/borderline resectable disease showed higher ORR and conversion rates, comparable to this borderline resectable group. Rudresha et al. reported a resectability rate of 23.8% after induction chemotherapy in patients with technically unresectable T4a OSCC20. The 32% achieving pCR (among oral cavity tumours resected) in our study suggests a strong potential for long-term survival benefits with induction chemoimmunotherapy in this subset. Patients with OSCC who undergo surgery after induction chemotherapy have improved 2-year OS compared to those who do not (47% vs. 20%)7.

Other studies have reported ORR of 30.5% after neoadjuvant TPF (Fu et al.) and 41.9% partial response in a retrospective audit (Mishra et al.)21,29 Noronha et al. in a prior study reported a 43% response rate with NACT in unresectable OSCC7. While our study showed a higher ORR of 78.2%, direct comparisons are challenging due to heterogeneous definitions and assessment methods for response across studies.

Phase 3 trials comparing IC with CRT have demonstrated ORRs of 77.7–80.1% with IC in unresectable disease30. Another trial in N2/N3 LAHNSCC reported ORR after IC of 64%31. However, both these trials did not show OS benefit with IC. The TAX 323 and TAX 324 trials reported ORRs of 54%/68% and 64%/72% with PF (cisplatin and 5-fluorouracil) and TPF (docetaxel, cisplatin, and 5-fluorouracil), respectively22,32,33. In comparison these trials, which primarily evaluated patients with resectable disease, our cohort consisted of predominantly unresectable HNSCC. Despite this, we achieved similar or higher ORR using a chemoimmunotherapy doublet/triplet regimen. This suggests that adding low-dose nivolumab might allow safe de-escalation from a triplet to a doublet regimen while maintaining high ORR and conversion rates.

Nivolumab as a single agent given for 3–4 cycles preoperatively has previously led to an ORR of 33%34. ICI combinations as NACT have shown lower response rates: the IMCISION trial reported a response in 6.25% of 32 patients treated with nivolumab ± ipilimumab35; nivolumab and lirilumab yielded an ORR of 11% among 28 patients36; neoadjuvant durvalumab with tremelimumab had an ORR of 43%37. These modest ORRs highlight the need for a chemotherapy backbone with ICIs to achieve high ORR and conversion to resectability.

In contrast, studies using chemotherapy with ICIs have yielded higher response rates. In the OPTIMA II Phase 2 trial of neoadjuvant nivolumab with chemotherapy for HPV-positive oropharyngeal cancer, deep responses (> 50% shrinkage) were observed in 70.8% of patients, and a pCR rate of 67% among those undergoing transoral robotic surgery15. A pilot phase 2 trial using three cycles of camrelizumab, nab-paclitaxel, and cisplatin reported an ORR of 89.6% in 48 patients, while Zhang et al. showed an impressive ORR of 96.7%, pCR of 37.0%, and major pathologic response of 74.1%, with camrelizumab-platinum-taxane in 30 patients with LAHNSCC38,39. Patel et al. reported ORR of 57% with neoadjuvant carboplatin, nab-paclitaxel, and durvalumab in resectable LAHNSCC40. At the time of writing, preliminary results from the phase III KEYNOTE-689 trial have been announced, marking the first demonstration of a significant event-free survival benefit (and a positive trend in OS) for perioperative immunotherapy in HNSCC. Although these initial findings are promising, we await the complete data presentation and peer-reviewed publication before considering any practice-changing implications41,42. It is important to note that trials investigating the addition of immunotherapy concurrent with chemoradiation did not show significant survival benefit43,44.

Compared to the aforementioned studies, our use of low-dose nivolumab plus chemotherapy in a mixed cohort of patients with unresectable/borderline resectable disease represents an innovative approach. We observed a higher ORR and conversion rates than those achieved with neoadjuvant ICI or induction chemotherapy alone, and mirrors the clinical benefit observed with standard-dose ICI plus chemotherapy.

A major barrier to ICI use across the world is the prohibitively high cost. Using low-dose nivolumab significantly reduced direct medical costs, making immunotherapy more broadly accessible. By reducing the cost by approximately 90%, we achieved substantial savings without compromising efficacy. This cost-effectiveness is crucial where patients pay out-of-pocket, highlighting a potential model to make innovative therapies accessible45. It should be noted that our estimates are based on current centre-specific pricing and reflect only the direct drug costs incurred by our patients. We did not incorporate variations in drug pricing or broader healthcare expenditures across different systems; this analysis is intended solely as a basic cost comparison of the dosing strategies at our centre.

Our findings have significant clinical and policy implications. Incorporating low-dose nivolumab into induction chemotherapy achieved a high ORR, converted a proportion of unresectable tumours to resectable status and had low additional toxicity. This approach could serve as a cost-effective, scalable model to improve access to ICIs not only in resource-constrained settings but also in high-income countries where optimizing cost and outcomes is increasingly important. Notably, a substantial proportion of our patients with baseline unresectable OCSCCs became eligible for surgery, which portends greatly improved prognosis and long-term survival.

The strengths of our study include the use of real-world data from a tumour board database and a novel chemoimmunotherapy regimen tailored to maximise cost-effectiveness. Our two-tiered follow-up minimized loss to follow-up and ensured accurate survival data.

The limitations of our study include potential selection bias due to the retrospective clinical data collection and individual patients’ ability to afford nivolumab. We did not have a control group which makes it difficult to attribute outcomes solely to the addition of nivolumab. While we attempted to harmonize the doses for analysis, variability in dosing frequency due to financial constraints introduces heterogeneity, potentially affecting results. Lastly, being a single-centre study limits generalizability.

By demonstrating that low doses of nivolumab can be both effective and economically advantageous, our study proposes a therapeutic strategy with broad applicability. This approach can be integrated into diverse healthcare systems, potentially benefiting patients with LAHNSCC worldwide by balancing efficacy with cost, regardless of the economic setting.

Our study supports further investigation of low-dose ICIs with chemotherapy in HNSCC. Future research should conduct prospective, randomised trials to validate these findings and determine the optimal nivolumab dosing strategy. A randomised trial studying low dose nivolumab with induction chemotherapy is currently accruing at our institution. We also suggest exploration of the biological mechanisms underlying the efficacy of low-dose ICIs and assess biomarkers that predict treatment response. Formal cost-effectiveness analyses are also needed to evaluate the economic impact of this treatment strategy in LMICs.

Conclusion

This ambispective real-world study demonstrates that the addition of low-dose nivolumab to induction chemotherapy in patients with locally advanced unresectable/borderline resectable non-metastatic HNSCC yields promising efficacy, as evidenced by a favourable overall response rate and a significant conversion to resectability in patients with oral cavity cancers and radical therapy for other sites. The regimen was generally well-tolerated and offers a cost-effective alternative, potentially broadening access to immunotherapy across diverse healthcare settings. While our findings are encouraging, further validation in randomised trials is warranted to confirm the efficacy, safety, and optimal dosing of low-dose ICI with induction chemotherapy in this patient population.