Flap sparing in postoperative radiotherapy versus standard flap-agnostic radiotherapy of oral cavity cancers (OPTIFLAP): protocol for a de-escalation, randomised, non-inferiority, phase III trial

Abstract

Introduction

The standard treatment of oral cavity cancers (OCC) relies on surgery and postoperative radiotherapy (poRT) for advanced stages or poor factors. In more than 75% of cases, reconstructive surgery with a flap aims to restore the function lost with tumour resection. Current poRT planning and delineation guidelines omit the presence of a flap. It may be assumed that poRT with flap sparing may allow for reducing radio-induced toxicities and improving functional outcomes, without impairing local primary control. The OPTIFLAP trial assesses non-inferior locoregional control using flap sparing compared with conventional flap-agnostic radiotherapy in patients with OCC, while reducing treatment-related toxicity and improving functional outcomes.

Methods and analysis

The OPTIFLAP study is a French, multicentre, 1:1 randomised, phase III, controlled trial. It will recruit 348 patients with OCC with a flap. Recruitment is active with the first enrolment on 2 July 2025 and is planned over 48 months. The primary outcome is non-inferior 2-year locoregional control rate using flap sparing compared with flap-agnostic radiotherapy (as per standard routine practice) in completely resected OCCs undergoing poRT. Key secondary outcomes include rates of toxicities, locoregional relapse-free survival, progression-free survival, overall survival, quality of life, functional outcomes (assessed by the Performance Status Scales for Head and Neck Cancer, the MD Anderson Dysphagia Inventory (self-questionnaire) and the Phonation Handicap Index (self-questionnaire)), flap doses and outcomes between arms depending on dosimetric parameters. The trial incorporates translational ancillary studies addressing individual radiosensitivity, salivary microbiome evolution, radiomics and dosiomics of flap changes, as well as medico-economic evaluation.

Ethics and dissemination

The study protocol has been approved by the Medical Ethics Committee East III (January 2025; Ref 24.05832.000442) and the French Agency for Medical and Health Products Safety (December 2024; ID-RCB: 2024-A01764-43) and was validated by review boards of all participating centres. Written informed consent will be obtained from all participants. Study results will be published in international peer-reviewed scientific journals and presented at relevant scientific conferences.

Trial registration number

NCT06798922.

Strengths and limitations of this study.

• This de-escalation interdisciplinary trial aims to optimise radiotherapy to make the most of reconstructive surgery with a flap in oral cavity cancer.

• It hypothesises that flaps could be partially spared from radiotherapy to limit their radiation-induced atrophy and fibrosis.

• It customises target volumes by irradiating the flap/native tissue junction rather than the full flap.

• Using a non-inferiority design on tumour control, the trial also aims at reducing radiation-induced flap damage.

• Small flaps, such as local flaps, cannot be spared due to uncertain target definition on imaging and technically limited radiotherapy modulation.

Introduction

Among head and neck cancers, the curative treatment of oral cavity cancers (OCCs) of advanced stages and those with positive margins relies on surgery and postoperative radiotherapy (poRT).¹ With advances in surgical instrumentation and optics, reconstructive surgery is performed in over 80% of cases, using a flap to restore the function lost with tumour resection. Such practice has been associated with good quality of life (QoL), good functional sequelae and good oncological outcomes.²

poRT worsens and accelerates atrophy and makes flaps fibrotic to the point where poRT alters the functional outcomes of patients with a flap.^{3 4} Current poRT planning and delineation guidelines omit the presence of a flap; radiation oncologists only recently had preliminary guidelines^{5 6} and an automatic segmentation tool of similar or better performance than expert flap delineation.^{7 8} Therefore, the poRT planning process as it is today may be suboptimal and should possibly account for the presence of a flap.

Preliminary radiotherapy data have shown that flaps are predominantly situated in the high-dose poRT area despite being ectopic non-tumorous tissues.⁸ These vascularised flaps, though essential for functional and aesthetic recovery, are not native tumour-bearing tissues. Current radiotherapy approaches do not differentiate between native and reconstructed tissue, leading to unnecessary irradiation of the flap. This can result in increased toxicity and compromised functional outcomes, including fibrosis, dysphagia and compromised phonation.

Emerging evidence suggests that sparing the flap from high-dose radiation, particularly its central components and osteotomised sections not involved in tumour spread, does not compromise oncologic safety. It may even improve postoperative function and QoL. Furthermore, advances in imaging and artificial intelligence (AI)-based segmentation tools have enabled more precise and reproducible delineation of flaps in radiotherapy planning.

Advances in surgical techniques may have some impact on the way and outcomes of radiotherapy. Beyond clinical discussions during tumour boards, it may be important to inform radiotherapy planning with a better understanding of technical developments.

Reconstructive surgery is a very active domain in head and neck surgery with continuous progress in flap rehabilitation and new biomaterials. Assuming that a flap is a non-tumorous tissue, only the tumour margins and junction between flap and native tissues may be considered at risk for cancer relapse.^{8 9} It may be assumed that poRT with flap sparing may allow similar local primary control while reducing radio-induced toxicities and improving functional outcomes.

Despite massive implementation of reconstructive surgery in OCCs, radiotherapy flap-sparing methods have been little integrated into routine practice due to a lack of understanding of radiation-induced flap changes and tools and fear of local relapse, notwithstanding the non-cancerous nature of flaps and location of a relapse risk mostly at the native tissue junction. Discussions between radiation oncologists and surgeons in front of the radiotherapy software should be encouraged to better handle the multidisciplinary issues of the management of flaps in radiotherapy planning and to optimise tumour volume definition. It is expected that follow-up personalisation and de-escalation of radiotherapy on flaps can reduce the morbidity of radiotherapy while ensuring similar local control rates. Radiation-induced flap changes include more substantial and accelerated atrophy and fibrosis compared with surgery alone, possibly osteoradionecrosis in bone flaps, and also larger volumes of irradiated normal tissues.

Fewer and less severe toxicities may also result in improved functional outcomes (swallowing, speech and cosmesis), more efficient rehabilitation and easier return to work and social interactions. It can also result in better prediction of definitive flap volume and function by surgeons at the time of flap harvesting (avoiding major overcompensation of flap volume and optimising flap selection), which is also relevant to donor site morbidity. For bone flaps, it might also contribute to better decisions for dental implants. The study should also help in understanding patterns of relapse and toxicities in the presence of different flaps and should be a starting point for the update of international guidelines for more accurate (voxel-scale) and personalised poRT.

In this context, and based on safety and efficacy data from our retrospective multicentric 400-patient XFLAP1 study,¹⁰ we proposed and implemented a de-escalation non-inferiority randomised trial, called OPTIFLAP, to compare current poRT practice without flap delineation and standard clinical target volume (CTV) definition and poRT with flap delineation. By challenging traditional radiotherapy volumes and integrating surgical anatomy into treatment planning, the study aims to refine multidisciplinary standards. If successful, flap-sparing radiotherapy could become a new standard, minimising toxicity, preserving long-term function and enhancing QoL for patients.

The OPTIFLAP trial addresses this unmet need by evaluating the oncologic safety and functional benefit of flap-sparing radiotherapy in a large, multicentric randomised trial with integrated translational research components.

Objective

The primary objective of the OPTIFLAP trial is to evaluate whether flap sparing yields a non-inferior 2-year locoregional control rate as flap-agnostic radiotherapy (as per standard routine practice) in completely resected OCCs undergoing poRT.

As the first secondary objective, the rate of ear-nose-throat (ENT) toxicities among patients with cavity cancer with a flap within 12 months after poRT will be compared between both strategies.

Other objectives include the comparison of the two strategies in terms of long-term toxicities (up to 5 years), clinical outcomes (including survival and implant complications), QoL, functional outcomes, cost-effectiveness and flap dosimetry correlation with toxicity.

Methods and analysis

Study design

OPTIFLAP is a French multicentric, open-label, non-inferiority controlled randomised phase III trial. It will enrol 348 patients from 23 private clinics/general hospitals/academic centres with histologically confirmed, resected oral cavity squamous cell carcinoma, who have undergone reconstructive flap surgery and are eligible for poRT. The OPTIFLAP trial is conducted in scientific collaboration with the Groupe d’Oncologie Radiothérapie Tête Et Cou (GORTEC)/Groupe d’Études de la Tête et du Cou (GETTEC) networks. It was presented to two scientific committees of the French Intergroup of Head and Neck Cancers, which brings together GORTEC/GETTEC, the French Rare Head and Neck Cancer Expertise Network, Unicancer, the Groupe Coopérateur Multidisciplinaire en Oncologie, as well as the Head and Neck Cancer patient association CORasso, for scientific validation and labelling. Patient recruitment and data collection started in July 2025. Inclusions are planned for 48 months. Patients will be followed up for 5 years, for an estimated end of data collection in July 2034.

The study protocol and this manuscript have been written in accordance with Standard Protocol Items for Randomised Trials.¹¹

Participants

Eligible patients are adults (≥18 years) with completely resected OCC, reconstructed using vascularised free or pedicled flaps and with an indication for poRT. Patients must have a performance status ≤2 at inclusion and may receive concomitant chemotherapy.

Patients are ineligible in the case of local flaps (not suitable for intensity-modulated radiotherapy), residual disease or distant metastases, prior head and neck radiotherapy, other recent malignancy, pregnancy or lactation or inability to comply with the study requirement in the opinion of the investigator.

Study procedures

Enrolment and randomisation

Eligible patients with OCC with an indication of poRT are proposed to participate after surgery by surgeons or radiation oncologists (figure 1). Patients are given study information and sufficient time to consider participation in the study. They provide a signed informed consent before any study procedure and are enrolled if they meet all eligibility criteria. Included patients are assigned by randomisation (ratio 1:1) between two strategies of poRT.

Figure 1. Study flowchart. CTV, clinical target volume; IMRT, intensity-modulated radiotherapy. Dose to CTV can vary from 50 to 66 Gy depending on histoprognostic risk factors, with doses per fraction of over 1,6 Gy / fracion in the low risk CTV.

Standard arm

Conventional flap-agnostic poRT, where flaps are included in the high-dose CTV following current international guidelines.

Experimental arm

Flap-sparing poRT, where non-involved flap tissue, including flap centres and osteotomies not at risk, are excluded from the high-dose CTV. Delineation follows updated consensus guidelines, preserving safety margins at the interface between native tissue and flap.

Randomisation is performed in the investigating centre using dedicated software via an internet portal. It is stratified on the following parameters: mobile tongue versus other locations; current presence of severe postsurgical locoregional complications (grade 3–4) versus absence and concomitant chemotherapy versus not.

poRT procedures

All patients are treated using image-guided, intensity-modulated radiotherapy (IMRT) or intensity-modulated proton therapy with a simultaneous integrated boost approach. Prescribed doses are aligned with French national guidelines and adapted per risk stratification (low, intermediate and high).

poRT will start within 4–9 weeks (no more than 10 weeks) following surgery. In the case of repeated surgery (carcinologic or flap problem), the delay is unchanged and still calculated in relation to the first surgery. Radiotherapy planning is based on CT without and with contrast enhancement.

Flap delineation follows the soft tissue flap delineation guidelines.⁵ The pedicled flaps may be delineated, distinguishing between the postoperative bed part and the inferior pedicle part. The bone flap component should be delineated separately from the native bone. It is also recommended to distinguish between the different osteotomies and to contour the fixation plates. As this is not current practice and dose constraints are lacking, these will be delineated post hoc anyway (based on individual case review images). The junction between the flap and the remaining native tissues should be delineated, with 5–6 mm expansion within the flap for the CTV, be it high, intermediate or low risk depending on the quality of resection and associated factors (perineural invasion, lymphatic and vascular emboli).^{6 8 10 12 13}

Median dose to the flap should be less than 40 Gy based on experts’ opinion and a small prospective clinical study⁹ but follows an ‘as low as reasonably achievable’ (ALARA) rule to not uncover the CTVs. It is recommended to maintain the maximal dose to the bone flap below 50 Gy, also according to the ALARA principle.

The flap microanastomotic area, or pedicle (rotated/twisted below tumour bed level), can be identified from the operative report, usually connecting the facial, lingual or superior thyroid artery to a flap artery. In pedicled flaps (also eligible in the trial), there is no microanatomotic area but instead a flap pedicle, i.e. the portion of the flap below the resected primary tumor area. Delineation should rely on a contrast-enhanced CT and is encouraged so as to avoid dose hot spots on vessels. The connected vessels at the recipient site can then be identified using radiologists’/surgeons’ expertise. As this is not current practice and dose constraints are lacking, these will be delineated post hoc anyway.

Maintaining the maximal dose to the vessels obeys the ALARA principle and can only be applied if the vessels are distant from any metastatic node, particularly if there is extracapsular spread.

For target volumes, the irradiation of the high-risk CTV (extracapsular extension and/or positive margins and/or tumorous bed) will be delivered daily 5 days a week, for a total dose of 60–66 Gy in 30–33 fractions (2 Gy/day). The prophylactic irradiation of low-risk CTV (non-clinically involved area corresponding to potential subclinical disease) will be delivered daily, 5 days a week, for a total dose of 50–54.45 Gy, in 30–33 fractions of 1.65 Gy/day (minimum 1.6 Gy/fraction, or higher upon physician’s appreciation; in the case where all CTVs are low risk, a total dose of 50 Gy in 25 fractions may be delivered).

Patients in the experimental arm will be treated by poRT with flap sparing, consisting of IMRT or proton therapy with dose painting on the junction area for flap sparing.⁹ In the standard arm, the definition of the target volumes should follow established guidelines14,16 but should further account for displacements/deformations of tissues following tumour resection and flap placement.

Withdrawal from study

The reasons why a patient may interrupt the treatment or discontinue participating in the study include the following circumstances:

• Disease progression.

• Unacceptable toxicity, not compatible with study treatment.

• Patient’s decision (the data already collected during the search can be kept and exploited unless the patient opposes it).

• Intercurrent illness or another reason that necessitates stopping treatment of the study.

• Patient lost to view.

• Investigator’s decision.

Any patient who prematurely withdraws from the study treatment only will continue to be followed, unless he/she withdraws from the study.

Data collection

Patients will be assessed up to 5 years after randomisation, as detailed in figure 2. Assessments include clinical examinations and toxicity assessments, tumorous evaluations, the completion of standardised self-questionnaires and biological samples. Data entry and handling are based on an electronic case report form. Patients are identified with a unique participant number. Access to the study database is restricted and secure.

Figure 2. Overview of study assessments. *Optional. **Centralised image review (during study and not before randomisation). ***Centralised dosimetry review: prospective individual case review of the first two patients in each centre before treatment and retrospective individual case review of all other cases (to be posted on a dedicated platform within 3 months of the patient’s radiotherapy completion). ATM, ataxia-telangiectasia mutated kinase; EORTC QLQ-C30, European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 developed to assess the quality of life of cancer patients in clinical trials; EQ5D5L, EuroQol 5-Dimension 5-Level questionnaire used in clinical and economic evaluations (including cost-utility analyses); GOHAI, Geriatric Oral Health Assessment Index; HN43, European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire – Head & Neck Cancer Module (43 items); MDADI, MD Anderson Dysphagia Inventory; PHI, Phonation Handicap Index; poRT, postoperative radiotherapy; VAS, visual analogue scale.

Study outcomes

Primary outcome

The primary endpoint is 2-year locoregional control, defined as the absence of tumour recurrence within the surgical bed or regional lymph nodes 2 years after randomisation, according to Response Evaluation Criteria in Solid Tumours (RECIST) 1.1 criteria. It is compared using flap-sparing compared with flap-agnostic radiotherapy (as per standard routine practice) in completely resected OCCs undergoing poRT.

Secondary outcomes

The first secondary outcome is the rate of ENT toxicities among patients with cavity cancer with a flap within 12 months after the end of poRT, defined as the proportion of patients with OCC with at least two ENT toxicities of grade ≥2 according to the National Cancer Institute-Common Terminology Criteria for Adverse Events (NCI-CTCAE) V.5.0 that occurred within 12 months after the end of poRT.

Clinically significant ENT toxicities include flap necrosis (total and partial), pharyngeal stenosis, neck fibrosis, poor tongue protraction (range of motion) reduction, trismus, osteoradionecrosis of flap or native bone, fistula, wound healing, atrophy and speech, as well as dysphagia, mucositis, xerostomia, dermatitis, etc.

Other secondary endpoints are:

• The proportion of patients with acute and late ENT toxicities evaluated according to NCI-CTCAE V.5.0, in terms of kind, grade, time of onset, reversibility and up to 5 years after the end of poRT.

• The proportion of patients with a local control rate up to 5 years after the end of poRT, according to the RECIST 1.1 criteria.

• The proportion of patients with implant complications, including subgroup analyses to take into account the type of flap.

• Locoregional relapse-free survival is defined as the time elapsed between randomisation and local recurrence (using RECIST 1.1 criteria) or death, whatever the cause.

• Progression-free survival (PFS) is defined as the time between randomisation and first progression according to RECIST 1.1 criteria or death, whatever the cause.

• Overall survival (OS) is defined as the time between randomisation and death, whatever the cause.

• Scores of QoL on standardised validated self-questionnaires (European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30 (EORTC-QLQ-C30) and its additional module European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire – Head & Neck Cancer Module (43 items) (HN43),¹⁷ Geriatric Oral Health Assessment Index,^{18 19} Visual Analogue Scale (VAS), EuroQol 5-Dimension 5-Level questionnaire (EQ5D5L)), functional assessments (Performance Status Scales for Head and Neck cancer, MD Anderson Dysphagia Inventory (MDADI)^{20 21} and Phonation Handicap Index (PHI)^{22 23} self-questionnaires).

• Differences in flap dose and outcomes between arms.

• Correlation between dose to the flap and toxicities (normal tissue complication probability modelling and ancillary study based on imaging by radiomics).

Translational research components

Through its integrated translational studies, OPTIFLAP will contribute to the understanding of tissue remodelling under radiation, the role of microbiota in mucosal resilience and the utility of AI for treatment personalisation. These insights could inform broader applications in head and neck oncology and reconstructive radiation planning.

Radiomics and dosiomics of flap evolution

Longitudinal imaging on follow-up will assess volumetric and textural changes in flap tissue using radiomic and dosiomic features. AI-driven tools (eg, 3D U-Net) enable segmentation and quantitative modelling of tissue remodelling, fibrosis and density evolution.

Radiosensitivity biomarkers

Ataxia-telangiectasia mutated kinase expression and phosphorylation levels will be assessed using immunohistochemistry and functional assays on patient-derived fibroblasts. The objective is to develop a radiosensitivity score correlating with normal tissue complications and flap-related outcomes.

Salivary microbiome profiling

Saliva samples collected at baseline and during follow-up will be analysed by 16S ribosomal RNA sequencing to evaluate the impact of radiation and flap exclusion on the oral microbiome. Correlations with mucosal toxicity and dysbiosis-associated symptoms will be explored.

Functional imaging and AI modelling

Changes in phonation and swallowing function will be assessed through video endoscopy, electromyography and AI-supported audio processing. Machine learning models will integrate imaging, treatment and clinical data to predict adverse functional outcomes.

Health economic evaluation

A cost-utility effectiveness analysis will be conducted from a societal and healthcare system perspective over a 5-year horizon, comparing flap-sparing to flap-agnostic radiotherapy in completely resected OCCs undergoing poRT. Direct medical costs, indirect costs and quality-adjusted life years will be collected and modelled using Markov simulations.

Statistical considerations

Sample size calculation

This phase III randomised study was designed as a non-inferiority study to observe non-inferior efficacy in terms of local disease control, with lower toxicities, in the experimental arm.

Based on our retrospective XFLAP1 study,¹⁰ in which we observed a 2-year local control rate of 82.4% in 247 patients with OCC reconstructed with flap and treated by radiotherapy, we hypothesise that the rate of 2-year local control in the OPTIFLAP study is expected to be 85% in the control group. Non-inferiority will be demonstrated if this rate is not statistically inferior to 75% in the experimental group. The demonstration of non-inferiority at a significance level of 0.05 (one-sided) with a power of 80% requires a sample size of 316 patients (158 per arm). Considering 10% of non-assessable patients, 348 will be included (174 per arm).

Furthermore, with the aim to conclude that toxicities will be lower in the experimental arm, the inclusion of 316 assessable patients will allow us to observe that the rate of severe ENT toxicities, expected as 15% in the experimental arm and 30% in the control arm, will be significantly different with alpha 5% and power of 89.4% for the bilateral test and of 94.1% for the unilateral test.

Statistical analysis

Since the trial is designed as a non-inferiority study, the primary analysis will be carried out by considering all assessable patients (per-protocol population), as this is the most conservative approach in this context. Patients will be considered as assessable if the treatment and the follow-up are compliant with the study protocol. A sensitivity analysis using the intent-to-treat population, considering all patients in their initial group of randomisation, will also be performed to test the robustness of the results.

The local control rate will be estimated with an associated 95% CI on patients with a minimum of 2 years of follow-up, unless local relapse is observed before, and non-inferiority will be assessed by the use of the z-test. The observed difference in the proportion of patients with local control at 2 years and its 95% unilateral CI will be calculated. If the unilateral CI does not include the 10% clinically relevant difference, then the de-escalation performed in the experimental group will be considered as non-inferior to the treatment of the control group.

To compare safety between arms, the rate of patients with ENT toxicities of grade ≥2 observed within 12 months after the end of radiotherapy will be compared between both arms by using the χ² test or Fisher’s exact test if necessary. Furthermore, acute and late ENT toxicities, evaluated according to NCI-CTCAE V.5.0, will be described up to 5 years after the end of poRT according to the kind, grade, time of onset and reversibility, and their proportions will be compared between the two arms using the χ² test or Fisher’s exact test, as appropriate.

The overall proportion of flap complications and the proportion of flap complications according to the type of flap will be compared between the two arms using the χ² test or Fisher’s exact test, as appropriate. Local relapse-free survival, PFS and OS curves will be calculated using the Kaplan-Meier method and compared between the two arms using the log-rank test. Scores of QoL on self-questionnaires (health-related QoL, QLQ-C30, HN43, MD Anderson Dysphagia Inventory (MDADI), Voice Handicap Index (VHI), Visual Analogue Scale (VAS) and EuroQol 5-Dimension 5-Level questionnaire (EQ5D5L)) will be described using mean, SD and range and will be compared between the two arms using Student’s t-test if the data follows a normal distribution or the Wilcoxon-Mann-Whitney test otherwise. Flap doses will be described and compared using Student’s t-test or the Wilcoxon-Mann-Whitney test, as appropriate. The correlation between the dose to the flap and toxicities will be measured using a one-way analysis of variance.

Quality assurance and data monitoring

The study conduction includes a centralised quality assurance programme of radiotherapy (RTQA) through the RTQA platform, ensuring protocol compliance through initial plan reviews (a benchmark case for the experimental arm and a dummy run case in the control arm), Digital Imaging and Communications in Medicine (DICOM, international standard for storing, transmitting, and sharing medical images and related information) data audits and contouring harmonisation workshops. A centralised review of dosimetry, imaging examinations and surgical and histological reports is planned for all randomised patients.

Data monitoring procedures are detailed in study-specific data management and trial monitoring plans. Data quality and compliance with the protocol and regulation are assessed throughout the trial by central monitoring and visits to sites. All serious adverse events are notified to and assessed by the sponsor and reported as part of safety annual reports.

Patient and public involvement

The CORasso patient association (https://corasso.org/) has been involved in the grant application, the protocol elaboration and the revision of the patient information file study protocol before regulatory procedures. This independent patient group will provide a patient perspective throughout the duration of the study.

Ethics and dissemination

This study received ethical approval from the Medical Ethics Committee East III in January 2025 (Reference 24.05832.000442) and from the French Agency for Medical and Health Products Safety in December 2024 (ID-RCB: 2024-A01764-43) and was validated by review boards of all participating centres.

The OPTIFLAP study has been registered under the identifier NCT06798922 on ClinicalTrials.gov from 23 January 2025. It will be conducted in accordance with the ethical standards of the Helsinki Declaration. All patients enrolled will give their written informed consent before study entry and any study procedures. The participant informed consent form in French can be found in the online supplemental material.

An individual participant data sharing statement is not planned. Study results will be published in international peer-reviewed scientific journals and presented at relevant scientific conferences.