Paclitaxel, carboplatin, and cetuximab induction chemotherapy for locally advanced head and neck squamous cell carcinoma: A retrospective evaluation of treatment outcomes and safety

Abstract

Background/Purpose

Definitive chemoradiotherapy (CRT) for locally advanced head and neck squamous cell carcinoma (HNSCC) has limitations in advanced disease. Paclitaxel, carboplatin, and cetuximab (PCE) offers a potentially safer induction chemotherapy alternative. This study evaluated the feasibility and outcomes of PCE induction followed by CRT compared with CRT alone among patients with Stage IV HNSCC who received definitive CRT.

Materials and methods

This retrospective study included 128 patients with Stage IV HNSCC without distant metastases treated at a single institution (PCE group, n = 52; CRT group, n = 76). Survival analyses were restricted to the 127 patients who initiated definitive CRT, with time measured from CRT initiation to reduce immortal time bias; CRT completion was analyzed as an outcome, not an eligibility criterion. An adjusted Cox model accounted for baseline imbalances in T classification and Stage.

Results

Median follow-up was 17.0 months (PCE) and 65.0 months (CRT). The PCE group had more advanced disease (T3-4: 90.4% vs. 65.8%; Stage IVB: 34.6% vs. 15.8%). PCE achieved an 82.7% response rate, CRT completion in 51 of 52 patients (98.1%), and significantly higher cisplatin ≥200 mg/m² achievement (86.5% vs. 59.2%, p = 0.002). PFS was longer with PCE in both unadjusted (HR: 0.515, 95% CI:0.271–0.980, p = 0.043) and adjusted analysis (adjusted HR: 0.477, 95% CI:0.247–0.924, p = 0.028). OS did not differ significantly (adjusted HR: 0.553, p = 0.147), and its interpretation is limited by the shorter follow-up in the PCE group. Exploratory inflammatory and nutritional markers showed no reliable prognostic association.

Conclusions

PCE induction was feasible, with a high rate of CRT completion and acceptable tolerability. PFS was longer in the PCE group after limited adjustment, whereas OS did not differ significantly. Given the retrospective, nonrandomized design, residual confounding, and shorter follow-up in the PCE group, these findings are hypothesis-generating and require confirmation in larger, prospective studies.

Introduction

Definitive chemoradiotherapy (CRT) for locally advanced head and neck squamous cell carcinoma (HNSCC) is widely accepted as a standard treatment that can achieve both organ preservation and favorable locoregional control.1, 2, 3 Meta-analyses have consistently demonstrated that the addition of concurrent chemotherapy significantly prolongs survival,² and its utility has been confirmed in pivotal trials with laryngeal preservation as the primary endpoint.³ However, in advanced Stage IV cases, distant metastasis and early recurrence limit the 5-year survival rate to 50-60%, highlighting the limitations of disease control achievable with CRT alone.

Induction chemotherapy (IC) has been explored as a strategy to address these challenges. IC aims to improve local control through tumor shrinkage, provide early intervention for micrometastases, and enable treatment stratification based on response. In regimen comparisons, docetaxel, cisplatin, and 5-fluorouracil combination (TPF) has been reported to improve progression-free survival (PFS) and overall survival (OS) compared to cisplatin and 5-FU(PF).^4,5 However, whether adding IC itself improves OS compared with CRT alone remains debatable, as major randomized controlled trials (such as the PARADIGM and DeCIDE trials) and integrated analyses have not demonstrated clear superiority, leaving the strategic role of IC under continued discussion.6, 7, 8, 9, 10

Furthermore, while TPF shows high response rates, it carries a substantial toxicity burden, with Grade 3-4 neutropenia occurring in 60-80% of cases and febrile neutropenia in 5-15%, often requiring hospitalization and treatment delays.^4,5 In addition, many studies have used the initiation of IC as the starting point for analysis, raising concerns about potential immortal time bias.¹¹

Given this background, there is a need to explore IC regimens that are more manageable in clinical practice while achieving both enhanced systemic control and high treatment completion. The paclitaxel, carboplatin, and cetuximab (PCE) regimen has garnered attention as a feasible option for real-world implementation owing to: ① its capability for outpatient administration, ② reduced renal burden through carboplatin use, and ③ milder myelosuppression compared with TPF.^12,13 In Japan, Enokida et al. reported a 2-year OS rate of 74.0% for PCE induction therapy followed by CRT in unresectable locally advanced HNSCC in a multicenter phase II trial (59 patients),¹² whereas Tahara et al. demonstrated a favorable overall response rate of 54.2% in a phase II trial for recurrent or metastatic cases.¹³ However, these studies have limitations, including: ① single-arm designs without direct comparison to a control group, ② potential immortal time bias with analysis starting at IC initiation,¹¹ and ③ insufficient detailed analysis of site-specific treatment effects.

The objective of this study was to retrospectively evaluate the treatment outcomes of patients with locally advanced HNSCC who received PCE IC followed by definitive CRT compared with those who received CRT alone. Importantly, to minimize immortal time bias, we set the analysis starting point at CRT initiation. The primary endpoints were OS and PFS, measured from CRT initiation, with secondary assessments of treatment completion rates, safety, and exploratory site-specific analyses. Through this approach, we aimed to clarify the clinical role of the sequential strategy of PCE induction followed by CRT and provide evidence to inform future treatment selection.

Methods and materials

Patient selection and study design

Patients with clinical Stage IV (IVA/IVB) HNSCC without distant metastases (M0) who were diagnosed and initiated treatment at Tokai University Hospital between January 2017 and December 2024 were included. Patients were assigned to either: 1) the PCE group, which received IC with the PCE regimen followed by CRT, or 2) the CRT group, which received CRT alone (Supplementary Figure S1). p16-positive oropharyngeal squamous cell carcinoma was excluded because it represents a distinct biological entity; under AJCC 8th edition staging, Stage IV p16-positive oropharyngeal cancer corresponds to distant metastatic disease (M1), which is not indicated for curative chemoradiotherapy. All 32 oropharyngeal cancer patients included in this study were p16-negative.

Patients eligible for PCE induction had T3 or higher disease, N2b or higher nodal involvement, or cases where tumor bulk was considered a barrier to effective radiotherapy. Treatment selection was made by the multidisciplinary tumor board based on clinical judgment rather than by randomization, so confounding by indication cannot be excluded. Of 133 patients screened, five who received another induction regimen (fluorouracil plus cisplatin) were excluded, leaving 128 patients for analysis (Supplementary Figure S1). One PCE patient who completed induction but died of sudden cardiac death before initiating CRT was excluded from the CRT-based survival analysis (n = 127).

Two analysis populations are distinguished throughout the manuscript. The treatment-delivery cohort comprises all 128 patients who received their assigned treatment and is used for baseline characteristics, induction response, safety, and CRT delivery. The CRT-initiated cohort comprises the 127 patients who began definitive CRT and is used for all survival analyses. CRT completion is reported as a treatment-delivery outcome and was not an eligibility criterion; patients who did not complete CRT were retained in both cohorts.

This analysis compares outcomes among patients who ultimately received definitive CRT, rather than a full intention-to-treat comparison from initial treatment decision.

Treatment

The PCE regimen consisted of paclitaxel 175 mg/m², carboplatin AUC 5, and cetuximab (400 mg/m² loading dose followed by 250 mg/m² weekly), administered for two cycles at 3-week intervals. Imaging assessments were performed after completion of PCE therapy, followed by the initiation of CRT.

CRT was delivered using intensity-modulated radiation therapy (IMRT) or volumetric modulated arc therapy (VMAT) at a total dose of 70 Gy in 35 fractions, with concurrent cisplatin 100 mg/m² administered every three weeks for three cycles. A 20% dose reduction was permitted based on the patient's body habitus and renal function.

Clinical and laboratory variables

Baseline data included age, sex, Eastern Cooperative Oncology Group Performance Status (ECOG-PS), clinical stage, and primary site. As an exploratory analysis, the following inflammatory and nutritional markers were derived from routine peripheral blood tests at two time points (before PCE induction and at CRT initiation): neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte-to-monocyte ratio (LMR), systemic immune-inflammation index (SII), prognostic nutritional index (PNI), C-reactive protein-to-albumin ratio (CAR), Glasgow Prognostic Score (GPS), and C-reactive protein (CRP). For each marker, an optimal cutoff was determined by receiver operating characteristic (ROC) analysis using the Youden index, and its association with progression-free survival was assessed by univariate Cox proportional hazards models.

Adverse events were assessed according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.¹⁴

Outcomes

OS was defined as the time from CRT initiation to death from any cause, while PFS was defined as the time from CRT initiation to disease progression or death. Tumor responses were evaluated using RECIST v1.1.¹⁵

Statistical analysis

For comparisons between groups, categorical variables were analyzed using the χ² test or Fisher's exact test, and continuous variables were analyzed using the Mann–Whitney U test. Survival curves were generated using the Kaplan–Meier method, and group comparisons were performed using log-rank tests.

Univariate Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI). The CRT group served as the reference. To address potential confounding from baseline imbalances, an adjusted Cox proportional hazards model was performed including treatment group, T classification (T3-4 vs T1-2), and Stage (IVB vs IVA) as covariates. These variables were selected a priori based on significant baseline differences between groups (Table 1). One patient in the PCE group who did not initiate CRT was excluded from this analysis (n = 127). A full multivariable model was not feasible due to the limited number of events relative to the number of potential covariates (generally ≥10 events per variable). This model provides limited adjustment for the measured imbalances in T classification and Stage; it does not eliminate residual confounding, and primary site, which also differed between groups, could not be included because of the small number of events.

Table 1.

Patient and tumor characteristics.

Characteristic	All (n = 128)		PCE (n = 52)		CRT (n = 76)		p-value	SMD
	n	%	n	%	n	%
Age, years							0.603	0.172
Median (range)	66	(33-84)	66	(33-80)	66	(36-84)
Sex							0.37	0.16
Male	101	78.9%	39	75.0%	62	81.6%
Female	27	21.1%	13	25.0%	14	18.4%
Smoking history							0.497	0.124
Yes	110	85.9%	46	88.5%	64	84.2%
No	18	14.1%	6	11.5%	12	15.8%
Alcohol use							0.678	0.075
Yes	96	75.0%	40	76.9%	56	73.7%
No	32	25.0%	12	23.1%	20	26.3%
Primary site							0.016	0.384
Nasopharynx	9	7.0%	1	1.9%	8	10.5%
Oropharynx	32	25.0%	16	30.8%	16	21.1%
Hypopharynx	50	39.1%	17	32.7%	33	43.4%
Larynx	12	9.4%	2	3.8%	10	13.2%
Sinonasal	19	14.8%	12	23.1%	7	9.2%
Other	6	4.7%	4	7.7%	2	2.6%
T classification							0.001	0.623
T1-2	31	24.2%	5	9.6%	26	34.2%
T3-4	97	75.8%	47	90.4%	50	65.8%
N classification							0.261	0.201
N0-1	35	27.3%	17	32.7%	18	23.7%
N2-3	93	72.7%	35	67.3%	58	76.3%
Stage							0.014	0.444
IVA	98	76.6%	34	65.4%	64	84.2%
IVB	30	23.4%	18	34.6%	12	15.8%

Baseline demographics and disease characteristics of the treatment-delivery cohort: patients receiving PCE induction chemotherapy followed by CRT (n = 52) versus CRT alone (n = 76). Groups were comparable in age, sex, performance status, and nodal classification. However, the PCE group had significantly more advanced T classification (T3-4: 90.4% vs. 65.8%, p = 0.001) and higher proportion of Stage IVB disease (34.6% vs. 15.8%, p = 0.014). Primary site distribution also differed significantly (p = 0.016), with the PCE group having more sinonasal cancers (23.1% vs. 9.2%).

CRT, chemoradiotherapy; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; PCE, paclitaxel, carboplatin, and cetuximab.

Statistical significance was set at a two-sided p < 0.05. Analyses were performed using EZR¹⁶ and Python 3.12 (lifelines 0.29.0).

Ethical considerations

This study was approved by the Ethics Review Board of Tokai University Hospital (approval number: 25R012) and conducted in accordance with the Declaration of Helsinki. The requirement for informed consent was waived, as this was a retrospective study using anonymized data.

Results

Patient characteristics

A total of 128 patients received their assigned treatment and formed the treatment-delivery cohort (PCE group, n = 52; CRT group, n = 76) (Supplementary Figure S1). Of these, 127 initiated definitive CRT and formed the CRT-initiated cohort used for all survival analyses (PCE, n = 51; CRT, n = 76). The median follow-up was 17.0 months (range: 0-55 months) for the PCE group and 65.0 months (range: 2-103 months) for the CRT group. Deaths occurred in eight of the 51 PCE patients who proceeded to definitive CRT (15.7%) and in 38 patients (50.0%) in the CRT group during the observation period.

Baseline characteristics are shown in Table 1. The median age was 66 years in both groups (p = 0.603). Most patients were male (75.0% PCE, 81.6% CRT, p = 0.370), with ECOG-PS 0-1 in most cases (p = 0.638).

Significant baseline imbalances were observed. Primary site distribution differed significantly (p = 0.016), with sinonasal cancers more frequent in the PCE group (23.1% vs 9.2%). The PCE group had more advanced T classification (T3-4: 90.4% vs 65.8%, p = 0.001) and higher Stage IVB proportion (34.6% vs 15.8%, p = 0.014).

Treatment response and selection for definitive treatment (CRT)

After completion of PCE therapy, the best response on computed tomography at the end of PCE induction (before CRT) was CR in three patients (5.8%), PR in 40 patients (76.9%), SD in six patients (11.5%), and PD in three patients (5.8%), with an overall response rate (ORR) of 82.7% and a disease control rate (DCR) of 94.2% (Table 2A). Fifty-one of the 52 patients in the PCE group (98.1%) proceeded to and completed definitive CRT; the remaining patient died of sudden cardiac death after induction, before CRT initiation.

Table 2.

Treatment response at the end of the PCE regimen and CRT.

(A) Best response after PCE induction chemotherapy (CT at end of PCE, before CRT) — PCE group, n=52
Site	n	CR	%	PR	%	SD	%	PD	%	ORR %	DCR %
All	52	3	5.8	40	76.9	6	11.5	3	5.8	82.7	94.2
Oropharynx	16	1	6.2	13	81.2	1	6.2	1	6.2	87.5	93.8
Hypopharynx	17	1	5.9	13	76.5	2	11.8	1	5.9	82.4	94.1
Nasopharynx	1	0	0	1	100	0	0	0	0	100	100
Larynx	2	0	0	2	100	0	0	0	0	100	100
Sinonasal	12	1	8.3	7	58.3	3	25	1	8.3	66.7	91.7
Other	4	0	0	4	100	0	0	0	0	100	100
(B) Final response after completion of CRT (CT/PET-CT) — both groups
PCE group (n=51; one patient who died during PCE before reaching CRT excluded)
Site	n	CR	%	PR	%	SD	%	PD	%	ORR %	DCR %
All	51	35	68.6	6	11.8	6	11.8	4	7.8	80.4	92.2
Oropharynx	16	12	75	0	0	4	25	0	0	75	100
Hypopharynx	16	12	75	1	6.2	2	12.5	1	6.2	81.2	93.8
Nasopharynx	1	1	100	0	0	0	0	0	0	100	100
Larynx	2	1	50	0	0	0	0	1	50	50	50
Sinonasal	12	8	66.7	3	25	0	0	1	8.3	91.7	91.7
Other	4	1	25	2	50	0	0	1	25	75	75
CRT group (n=76)
Site	n	CR	%	PR	%	SD	%	PD	%	ORR %	DCR %
All	75	59	78.7	4	5.3	4	5.3	8	10.7	84	89.3
Oropharynx	15	11	73.3	2	13.3	1	6.7	1	6.7	86.7	93.3
Hypopharynx	33	25	75.8	2	6.1	2	6.1	4	12.1	81.8	87.9
Nasopharynx	8	7	87.5	0	0	0	0	1	12.5	87.5	87.5
Larynx	10	9	90	0	0	0	0	1	10	90	90
Sinonasal	7	5	71.4	0	0	1	14.3	1	14.3	71.4	85.7
Other	2	2	100	0	0	0	0	0	0	100	100

Treatment response evaluated by RECIST version 1.1. Section A shows the best response after PCE induction chemotherapy, assessed by computed tomography at the end of PCE (before CRT) (PCE group, n = 52): overall response rate (ORR) was 82.7% and disease control rate (DCR) was 94.2%. Section B presents the final response after completion of CRT, assessed by computed tomography and PET-CT, for both groups: the PCE group that proceeded to CRT (n = 51) achieved an ORR of 80.4% and DCR of 92.2%; the CRT group achieved an ORR of 84.0% and DCR of 89.3% among 75 evaluable patients (one of the 76 patients was not evaluable). Site-specific responses are detailed by oropharynx, hypopharynx, larynx, and sinonasal subsites. Sinonasal cancers showed high response rates in the PCE group (91.7% ORR at CRT completion).

CR, complete response; CRT, chemoradiotherapy; DCR, disease control rate; ORR, overall response rate; PCE, paclitaxel, carboplatin, and cetuximab; PD, progressive disease; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors; SD, stable disease.

At CRT completion, the ORR was 80.4% and the DCR was 92.2% in the PCE group that proceeded to CRT (n = 51; Table 2B). In the PCE group, the ORRs were 75.0%, 81.2%, and 91.7% for oropharyngeal, hypopharyngeal, and sinonasal cancers, respectively. In the CRT group, the ORR was 84.0% and the DCR was 89.3% among the 75 evaluable patients (one of the 76 patients was not evaluable).

Cisplatin ≥200 mg/m² during CRT was achieved in 45 of 52 patients (86.5%) in the PCE group compared with 45 of 76 patients (59.2%) in the CRT group (p = 0.002; Table 4).

Table 4.

Chemoradiotherapy tolerability by treatment group.

	PCE + CRT (n = 52)		CRT alone (n = 76)		p-value
	n	%	n	%
CRT completion	51	98.1%	67	88.2%	0.057
Cisplatin ≥200 mg/m² achieved	45	86.5%	45	59.2%	0.002

Tolerability of definitive chemoradiotherapy in the two groups. CRT completion and achievement of a cumulative cisplatin dose of ≥200 mg/m² are shown; denominators are those of the treatment-delivery cohort (PCE, n = 52; CRT, n = 76). The PCE group achieved a higher CRT completion rate (98.1% vs. 88.2%, p = 0.057) and a significantly higher rate of cisplatin ≥200 mg/m² (86.5% vs. 59.2%, p = 0.002). The single PCE patient counted as not completing CRT died of sudden cardiac death after induction, before CRT initiation; all 51 PCE patients who initiated CRT completed it. Detailed comparative toxicity during CRT was not available, so these data describe treatment delivery rather than a full toxicity comparison.

CRT, chemoradiotherapy; PCE, paclitaxel, carboplatin, and cetuximab.

Survival outcomes

Overall survival analysis

The median OS was not reached in the PCE group versus 47 months in the CRT group. The 2-year OS rates were 75.6% (PCE) and 66.6% (CRT). The median PFS was not reached in the PCE group versus 22 months in the CRT group. The 2-year PFS rates were 74.1% (PCE) and 49.4% (CRT). Because median follow-up was shorter in the PCE group (17.0 vs 65.0 months), estimates beyond 2 years in that group are based on few patients at risk, and the median OS was not reached partly for this reason. Kaplan-Meier curves are shown in Fig. 1, Fig. 2.

Fig. 1.

Overall Survival in Patients with Locally Advanced Head and Neck Squamous Cell Carcinoma.

Kaplan-Meier curves for overall survival (OS). Panel A shows the CRT-initiated cohort (n = 127). Panel B presents OS comparison between PCE+CRT (n = 51) and CRT alone (n = 76) (HR 0.585, 95% CI 0.270–1.270, p = 0.176). The difference was not statistically significant. Panels C–E display exploratory subgroup analyses by primary tumor site. Numbers at risk are provided. Analysis started at CRT initiation to reduce immortal time bias. Median follow-up: PCE 17.0 months, CRT 65.0 months.

CI, confidence interval; CRT, chemoradiotherapy; HR, hazard ratio; OS, overall survival; PCE, paclitaxel, carboplatin, and cetuximab.

Fig. 2.

Progression-Free Survival in Patients with Locally Advanced Head and Neck Squamous Cell Carcinoma.

Kaplan-Meier curves for progression-free survival (PFS). Panel A shows the CRT-initiated cohort (n = 127). Panel B presents PFS comparison between PCE+CRT (n = 51) and CRT alone (n = 76) (HR 0.515, 95% CI 0.271–0.980, p = 0.043). After adjustment for T classification and Stage, PFS remained significant (adjusted HR 0.477, 95% CI 0.247–0.924, p = 0.028; Table 3C). Panels C–E display exploratory subgroup analyses by primary tumor site. Numbers at risk are provided. Median follow-up: PCE 17.0 months, CRT 65.0 months.

CI, confidence interval; CRT, chemoradiotherapy; HR, hazard ratio; PFS, progression-free survival; PCE, paclitaxel, carboplatin, and cetuximab.

Univariate analysis of prognostic factors

The results of the univariate Cox proportional hazards regression analysis are presented in Tables 3.

In univariate analysis, PFS was significantly improved with PCE induction (HR: 0.515, 95% CI: 0.271–0.980, p = 0.043; Fig. 2). OS showed a non-significant trend favoring PCE (HR: 0.585, 95% CI: 0.270–1.270, p = 0.176; Fig. 1). Median OS was not reached in the PCE group versus 47 months in the CRT group. Median PFS was not reached in the PCE group versus 22 months in the CRT group. (Table 3).

Table 3.

Cox regression analysis for overall survival and progression-free survival.

(A) Overall Survival — Univariate Analysis
Variable	Hazard Ratio (95% CI)	P-value	n
Treatment group (PCE vs CRT)	0.585 (0.270–1.270)	0.176	127
Age	1.009 (0.976–1.044)	0.582	127
T classification (T3-4 vs T1-2)	0.838 (0.451–1.556)	0.575	127
N classification (N+ vs N0)	1.110 (0.517–2.381)	0.789	127
Stage (IVB vs IVA)	1.222 (0.620–2.408)	0.562	127
(B) Progression-Free Survival — Univariate Analysis
Variable	Hazard Ratio (95% CI)	P-value	n
Treatment group (PCE vs CRT)	0.515 (0.271–0.980)	0.043	127
Age	1.001 (0.974–1.028)	0.963	127
T classification (T3-4 vs T1-2)	1.014 (0.571–1.803)	0.961	127
N classification (N+ vs N0)	1.097 (0.582–2.067)	0.774	127
Stage (IVB vs IVA)	1.162 (0.647–2.089)	0.616	127
(C) Adjusted Cox Proportional Hazards Analysis†
Overall Survival
Variable	Hazard Ratio (95% CI)	P-value	n
Treatment group (PCE vs CRT)	0.553 (0.248–1.231)	0.147	127‡
T classification (T3-4 vs T1-2)	0.889 (0.475–1.664)	0.713
Stage (IVB vs IVA)	1.407 (0.700–2.830)	0.338
Progression-Free Survival
Variable	Hazard Ratio (95% CI)	P-value	n
Treatment group (PCE vs CRT)	0.477 (0.247–0.924)	0.028	127‡
T classification (T3-4 vs T1-2)	1.133 (0.630–2.036)	0.677
Stage (IVB vs IVA)	1.315 (0.722–2.396)	0.371

Sections A and B present univariate Cox proportional hazards analysis for overall survival and progression-free survival, respectively. Section C presents the adjusted Cox model including treatment group, T classification (T3-4 vs T1-2), and Stage (IVB vs IVA) as covariates. PFS was significantly improved with PCE in both unadjusted (HR 0.515, p = 0.043) and adjusted analysis (adjusted HR 0.477, p = 0.028). OS showed a non-significant trend (adjusted HR 0.553, p = 0.147). One patient who did not initiate CRT was excluded from the adjusted analysis (n = 127). Section C provides limited adjustment for measured imbalances and does not eliminate residual confounding.

CI, confidence interval; CRT, chemoradiotherapy; HR, hazard ratio; OS, overall survival; PCE, paclitaxel, carboplatin, and cetuximab; PFS, progression-free survival.

^†Adjusted for T classification (T3-4 vs T1-2) and Stage (IVB vs IVA) based on significant baseline imbalances.

^‡One patient in the PCE group who did not initiate CRT was excluded from the adjusted analysis.

Subgroup analysis by primary site

Subgroup analysis according to the major primary sites (oropharynx, hypopharynx, and sinonasal) is shown in Supplementary Table S1. These analyses were performed within the CRT-initiated cohort (n = 127); the hypopharyngeal subgroup therefore comprises 49 of the 50 patients with hypopharyngeal cancer in the treatment-delivery cohort.

For oropharyngeal cancer (n = 32), the 2-year OS rates were 81.5% in the PCE group and 75.0% in the CRT group, with no significant difference (p = 0.506).

For hypopharyngeal cancer (n = 49), the 2-year OS rates were 63.2% in the PCE group and 62.5% in the CRT group, with no significant difference (p = 0.649).

In sinonasal cancer (n = 19), the PCE group showed higher survival estimates: 2-year OS 88.9% versus 42.9% (p = 0.145), with the same estimates at 3 years. Two-year PFS was 90.9% versus 42.9% (p = 0.044). Because this subgroup was small (PCE, n = 12; CRT, n = 7), included only one event in the PCE group, and was not adjusted for multiplicity, these results are exploratory. Survival curves stratified by primary site are shown in Fig. 1, Fig. 2.

Adjusted cox proportional hazards analysis

After adjustment for T classification and Stage (Table 3C), PFS was statistically significant (adjusted HR: 0.477, 95% CI: 0.247–0.924, p = 0.028). OS showed a non-significant trend (adjusted HR: 0.553, 95% CI: 0.248–1.231, p = 0.147). T classification and Stage were not independently significant in the adjusted model. This model provides limited adjustment rather than definitive control of confounding, and residual confounding from unmeasured factors, including primary site, remains possible.

Exploratory analysis of inflammatory and nutritional markers

The inflammatory and nutritional markers described in the Methods were examined in the PCE group as potential prognostic factors. No marker was identified as a reliable prognostic factor: the associations were inconsistent in direction, were based on a small number of events, and used cutoff values derived from the same dataset without correction for multiple comparisons. These exploratory analyses are reported only to indicate that no clinically useful biomarker emerged in this cohort, and no further conclusions are drawn from them.

Safety

Grade 3 or higher adverse events from PCE IC occurred in 6/52 patients (11.5%)(Supplementary Table S2). These included neutropenia in four patients (7.7%) and hypomagnesemia in two patients (3.8%). No Grade 4 or 5 events occurred. The most common adverse events were alopecia (86.5%; predominantly Grade 2), acneiform rash (75.0%; predominantly Grade 1), and hand-foot syndrome (55.8%; predominantly Grade 1).

No delays occurred in the initiation of definitive CRT. PCE therapy was completed in all patients, and cisplatin ≥200 mg/m² during definitive CRT was achieved in 45 of 52 patients (86.5%).

Discussion

This study retrospectively evaluated survival outcomes from CRT initiation in patients with American Joint Committee on Cancer (AJCC) 8th edition Stage IVA–IVB HNSCC who received PCE IC followed by CRT.¹⁷ The main findings were as follows: ① PFS was longer with PCE induction in both unadjusted (HR: 0.515, p = 0.043) and adjusted analysis (adjusted HR: 0.477, p = 0.028), while OS showed a non-significant trend (adjusted HR: 0.553, p = 0.147). PCE therapy was associated with a high CRT completion rate of 98.1%, with Grade 3 or higher adverse events occurring in 11.5% of the patients. In an exploratory subgroup analysis by primary site, the PCE group showed higher survival estimates for sinonasal cancer (2-year OS: 88.9% vs. 42.9%, p = 0.145).

Significance of PCE therapy

The significance of IC for locally advanced head and neck cancer has been debated in several trials. While TAX 323¹⁸ and TAX 324⁴ demonstrated TPF superiority over PF, the DeCIDE⁷ and PARADIGM⁶ trials failed to demonstrate clear survival benefits, raising concerns about treatment-related toxicity and completion rates.

There is a need to explore IC regimens that achieve both systemic control and high treatment completion. PCE therapy integrates three mechanisms: systemic control (carboplatin), radiosensitization (paclitaxel),¹⁹ and molecular targeting (cetuximab),²⁰ while offering outpatient administration convenience.

The 2-year OS rate of 75.6% in our PCE group is consistent with Shirasu et al.,²¹ and supports the feasibility of PCE in real-world settings. The high rate of CRT completion (98.1%) is consistent with a manageable induction regimen, although no direct comparison with TPF was made in this study. Current Japanese guidelines²² recommend only TPF (Grade B1) and FP (Grade C1) as IC regimens; our findings suggest that PCE warrants prospective evaluation as an alternative.

Considerations regarding failure to reach statistical significance

In this study, PFS was significantly improved with PCE induction after adjustment for baseline imbalances (adjusted HR: 0.477, p = 0.028), while OS did not reach statistical significance (adjusted HR: 0.553, p = 0.147). Several factors may explain the non-significant OS finding: (a) Insufficient statistical power due to limited events (8 deaths in the PCE group); (b) Short median follow-up in the PCE group (17.0 months) with immature survival curves; (c) Baseline imbalance with the PCE group having more advanced disease (T3-4: 90.4% vs. 65.8%, p = 0.001; Stage IVB: 34.6% vs. 15.8%, p = 0.014).

To address confounding from baseline imbalances, a parsimonious adjusted Cox model including T classification and Stage as covariates was performed. PFS was significantly improved after adjustment (adjusted HR: 0.477, 95% CI: 0.247–0.924, p = 0.028), while OS showed a non-significant trend (adjusted HR: 0.553, p = 0.147). The shorter median follow-up in the PCE group (17.0 vs. 65.0 months) limits comparison of long-term OS. A full multivariable model was not feasible due to limited events per variable.^23,24

Despite more advanced baseline disease, survival in the PCE group was at least comparable to that in the CRT group. This observation is compatible with, but does not establish, an effect of PCE induction on tumor control. A meta-analysis by Pignon et al.,²⁵ reported that IC might have greater effects in advanced cases. Prospective randomized studies with balanced baseline characteristics are needed to determine whether a survival benefit exists.

Favorable results in sinonasal cancer

In an exploratory subgroup analysis, the PCE group showed higher survival estimates in sinonasal cancer (2-year OS: 88.9% vs. 42.9%, p = 0.145; 2-year PFS: 90.9% vs. 42.9%, p = 0.044). Several factors could contribute to this observation, although all remain speculative.

Sinonasal cancers are often difficult to resect and are subject to radiation dose constraints imposed by adjacent critical organs²⁶; tumor shrinkage after induction may relax these constraints.²⁷ Sinonasal squamous cell carcinoma has also been reported to be relatively chemosensitive,²⁸ and in the present cohort the disease control rate at induction was 91.7%, with deepening of responses after CRT (CR in eight of 12 patients). Effects of induction chemotherapy on the tumor microenvironment, including vascular normalization, have been proposed as a further mechanism.^29,30

However, this subgroup analysis included only 19 patients (PCE, n = 12; CRT, n = 7) with a single death in the PCE group, was not adjusted for multiplicity, and had insufficient statistical power. This finding is hypothesis-generating only and requires validation through prospective studies focusing on sinonasal cancers.

Reasons for limited efficacy in hypopharyngeal cancer

In the hypopharyngeal cancer subgroup, no clear difference was observed between the two groups (2-year OS: 63.2% vs. 62.5%, p = 0.649). Hypopharyngeal cancer has one of the poorest prognoses among head and neck cancers,³¹ and the following characteristics may contribute to treatment resistance:

• Extensive submucosal spread and risk of extracapsular extension due to rich lymphatic drainage³²;• Vulnerability in treatment completion associated with dysphagia, malnutrition, and lifestyle factors (smoking/alcohol)³³; • Reduced sensitivity to drugs and radiation due to biological characteristics such as tumor hypoxia³⁴; • Difficulty in local control due to large irradiation volumes and dose constraints.

In hypopharyngeal cancer, long-term control may be difficult due to these biological characteristics, even if temporary shrinkage is achieved with IC. For this site, more intensive induction therapy or treatment-intensity adjustments based on post-induction responses (response-adapted strategies) may be necessary. Haddad et al. reported the potential for de-escalation strategies in patients with favorable responses to IC,⁶ highlighting the importance of individualized treatment.

Study limitations

This study has the following important limitations:

Study design limitations: This single-center retrospective study had non-standardized treatment selection and potential selection bias. Because PCE induction was selected by the multidisciplinary tumor board on clinical grounds rather than by randomization, confounding by indication cannot be excluded. The PCE group included more patients with advanced disease, and treatment protocols may have differed across the 2017-2024 observation period.

Statistical limitations: Limited event numbers (8 deaths, 12 progressions in the PCE group) limited the adjusted model to two covariates; a full multivariable analysis was not feasible. The median follow-up for the PCE group (17.0 months) was shorter than the CRT group (65.0 months), limiting long-term OS assessment. Multiple site-specific comparisons were performed without multiplicity adjustment; sinonasal cancer results should be interpreted as exploratory. Similarly, the exploratory analysis of inflammatory and nutritional markers did not identify any reliable prognostic factor and should be regarded as hypothesis-generating only.

Data limitations: Important prognostic factors such as tumor volume, comorbidity indices, and renal function were not available in the retrospective dataset. The imbalances in T classification and Stage were adjusted for in a parsimonious Cox model, but this represents limited adjustment rather than definitive control of confounding; primary site remained imbalanced, with sinonasal cancers more frequent in the PCE group, and residual confounding from unmeasured variables cannot be excluded.

Absence of quality of life (QOL) data: Longitudinal evaluations of swallowing function, voice function, and QOL were not performed.

Cost-effectiveness not evaluated: A health-economic evaluation of PCE therapy was not conducted.

Clinical implications

Despite the above limitations, this study has the following clinical implications:

First, PCE therapy was associated with a high CRT completion rate (98.1%) and an acceptable safety profile during induction (Grade 3 or higher adverse events, 11.5%). This indicates that PCE therapy is a feasible IC option in real-world practice. Major trials of TPF therapy have reported high incidences of severe hematologic toxicity and infections⁵; however, the two regimens were not compared directly in this study. Additionally, outpatient administration reduces patient burden.

Second, PFS was longer in the PCE group after limited adjustment for baseline imbalances (adjusted HR: 0.477, p = 0.028), and OS showed a non-significant trend despite more advanced baseline characteristics. These findings are hypothesis-generating and do not establish comparative efficacy.

Third, the exploratory sinonasal findings, based on 19 patients, should be regarded as hypothesis-generating and require prospective validation.

Fourth, current treatment guidelines in Japan and the United States^35,36 recommend only TPF therapy (recommendation Grade B1) and FP therapy (recommendation Grade C1) as IC regimens. This study suggests that PCE merits further prospective investigation as an alternative induction regimen.

Conclusion

In this retrospective analysis of Stage IV HNSCC, PCE induction chemotherapy was associated with a high rate of CRT completion (98.1%) and acceptable tolerability during induction. PFS was longer in the PCE group in both unadjusted (HR: 0.515, p = 0.043) and limited adjusted analysis (adjusted HR: 0.477, p = 0.028), whereas OS did not differ significantly. An exploratory subgroup analysis suggested more favorable outcomes in sinonasal cancer. Because of the retrospective, nonrandomized design, residual confounding, and the shorter follow-up in the PCE group, these findings are hypothesis-generating and require confirmation in larger, prospective studies.

Supplementary Figure S1. patient selection flowchart

Flow diagram of patient selection. Between January 2017 and December 2024, 133 patients with clinical Stage IV (no distant metastases, M0) head and neck squamous cell carcinoma were treated at Tokai University Hospital. Five patients who received an induction chemotherapy regimen other than PCE were excluded, leaving 128 patients in the treatment-delivery cohort: 76 in the CRT group (CRT alone; IMRT 70 Gy/35 fractions with cisplatin) and 52 in the PCE group. Induction chemotherapy was indicated for patients with T3 or higher and/or N2b or higher disease. Of the 52 patients in the PCE group, 51 proceeded to definitive CRT; one patient died of sudden cardiac death after completing induction, before CRT initiation. The CRT-initiated cohort used for survival analyses therefore comprised 127 patients.

CRT, chemoradiotherapy; HNSCC, head and neck squamous cell carcinoma; PCE, paclitaxel, carboplatin, and cetuximab.

Specific declarations

Financial interests

None of the authors have received any financial support, grants, or honoraria related to this research from pharmaceutical companies, device manufacturers, or other commercial entities.

Research support

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. The study was conducted using institutional resources at Tokai University Hospital.

Employment

All authors are employed by Tokai University School of Medicine and Tokai University Hospital. No author has any employment relationship with entities that could benefit from the results of this study.

Consultancy

No author has served as a consultant for any pharmaceutical company or medical device manufacturer related to the treatments or products discussed in this manuscript.

Stock/Shares

No author owns stocks or shares in any company that could benefit from the publication of this research.

Patents

No author has filed or holds any patents related to the content of this manuscript.

Personal relationships

There are no personal relationships among the authors or with others that could inappropriately influence this work.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix. Supplementary materials

Supplementary Table S1. Subgroup Analysis Details by Primary Site for Overall Survival

Supplementary Table S2. Detailed Safety Profile of PCE Regimen