Photobiomodulation in the management of oral mucositis for adult head and neck cancer patients receiving irradiation: the LiTEFORM RCT

Description of the condition

Oral mucositis (OM) has the unfortunate distinctions of being both the most common and the most debilitating complication of head and neck cancer (HNC) irradiation. 1 It is characterised by inflammation of the mucous membranes with erythema and ulceration (Figure 1).

FIGURE 1.

Example of oral mucositis on the tongue.

Many patients need opioid medications to control the pain, which affects their ability and willingness to eat and drink throughout radiotherapy and, for some, months thereafter. Consequently, around 90% of patients will require nutritional support and most will require tube feeding.

Feeding tubes can be inserted through the nose (Figure 2) or directly through the abdominal wall into the stomach. 2 The pain of OM for patients is only the beginning of the story. Patients’ general, psychosocial and financial well-being, and that of their carers, are also affected by OM. 3 Mucositis is an independent risk factor for pharyngo-oesophageal stricture. Stricture is a devastating complication, which can develop after HNC radiotherapy as a result of scarring partially or completely blocking the gullet and can result in a permanent inability to swallow, an aspiration of food and fluid into the lungs, and a long-term dependence on feeding tubes. 4 Strictures typically present within the first 6 months following the completion of irradiation, but can develop up to 5 years following completion. 5

FIGURE 2.

Example of a patient with a nasal feeding tube. Used with permission from the photo subject.

Size of the problem

Six thousand patients per year in England and Wales undergo (chemo)radiotherapy [(C)RT] for HNC. 6 Ninety-seven per cent of these patients will develop OM because of their (C)RT. Patients receiving (C)RT are at greater risk of more severe OM than those who are not because of potentiation of the effects of the radiotherapy by the chemotherapy. 7

How does oral mucositis develop?

The natural history of OM is gradually being unravelled. In 2004, Sonis et al. 3 described a five-stage mechanism by which the condition develops and then heals (Figure 3). These stages are initiation, the primary damage response, signalling and amplification, ulceration, and healing. In the first stage (initiation), the mucosal cells are injured by the cancer therapy. In the next two stages, reactive oxygen species and inflammatory cytokines (released by direct tissue injury) cause further damage to the submucosa, leading to ulceration (stage 4). Microbial toxins are also thought to stimulate further inflammation of the ulcerated lesion. In stage 5, which is the least well-understood phase, healing takes place. In the UK, cancer treatments are delivered daily from Monday to Friday, but not at weekends or bank holidays. Radiotherapy continues for 6–7 weeks, much longer than the 15–21 days in Sonis’s table. 3 Concurrent chemotherapy is given either weekly or every 3 weeks for the duration of the radiotherapy. 8 In effect, all five stages may be happening at the same time.

FIGURE 3.

Sonis et al. ’s3 five stages of mucositis. Reproduced with permission from Basile et al. 9 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Basile et al. 9 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Current oral mucositis treatment strategies in the UK

In general, patients are encouraged to improve oral hygiene as a preventative measure. Mouthwashes are typically prescribed in an effort to protect the oral mucosa, by keeping the oral mucosa moist and clean during treatment. Mouthwashes, such as Biotène™ (GlaxoSmithKline plc, Brentford, UK), also help with the side effect of mouth dryness. A variety of mouthwashes are used, but there is no evidence of any of them being effective in relation to OM. It is suggested that chlorhexidine mouthwash is not used because its alcohol content makes it painful to use. 10

Nutrition is optimised by prescribing supplement drinks for patients who find chewing solid foods difficult and for those whose oral intake remains inadequate despite dietary changes to softer or liquid-textured foods. Pain is managed using combinations of topical and systemic analgesics, following the World Health Organization (WHO)’s analgesia ladder11 as symptoms progress. Anecdotal evidence suggests that there remains a variation in management strategies in the UK.

In terms of the process described by Sonis et al. 3, it is possible that stage 4 is suppressed by these measures. Mouthwash irrigation may reduce the bacterial load in the ulcers, and coating gels may cover mucosal breaches, reducing discomfort. However, none of the current approaches has the ability to affect the progression through the stages.

Rationale for the LiTEFORM trial

Low-level laser therapy (LLLT), or photobiomodulation, is a treatment that has the potential to reduce the severity of OM. LLLT involves the application of a low-powered laser to the affected tissue. The most familiar types of medical lasers are those that are used to cut or ablate tissues. LLLT works in dramatic contrast to this, modulating biochemical pathways within the cell to reduce inflammation and improve healing. The mechanism by which this happens is not fully understood. It is postulated that the light is absorbed into the mitochondria, which increases the activity of the cell, accelerates cell healing and inhibits pain. 12 It is plausible that LLLT is modulating all five stages of Sonis et al. ’s3 mucositis model. The effect of the laser depends on the wavelength and density of the light, as well as the duration and frequency of application. The timing of LLLT with respect to radiotherapy treatment sessions may also be significant.

Evidence for the use of low-level laser therapy

Low-level laser treatment for OM in HNC is gaining popularity outside the UK. 13 Results from trials have been encouraging. There has been a series of systematic reviews conducted over the last decade, including two by the Cochrane collaboration. 14 These reviews have pointed towards progressively stronger evidence supporting the use of low-level lasers for managing OM in HNC.

To our knowledge, the most up-to-date systematic review prior to starting the LiTEFORM trial was conducted by Oberoi et al. 15 This included 18 randomised controlled trials (RCTs) of LLLT for OM, 10 of which related to patients with conditions other than HNC who were treated with a mixture of chemotherapy and radiotherapy. This review concluded that prophylactic LLLT reduced severe OM in patients with cancer [risk ratio 0.37, 95% confidence interval (CI) 0.20 to 0.67; p = 0.001]. Many of these patients did not receive radiotherapy but experienced OM as a result of high-dose chemotherapy. The review suggested that future research should identify the optimal characteristics of LLLT and determine the feasibility of using LLLT in the clinical setting.

Allocation concealment

Published LLLT trials have varied methods for delivering low-level laser treatment. The method of blinding both staff and patients to laser treatment tended to be poorly described, affecting the trial results owing to potential reporter bias. In Oberoi et al. ’s15 review, it was noted that only 21% of studies reported adequate allocation concealment. Furthermore, none of the studies included in the systematic reviews attempted to blind the clinicians delivering the LLLT. Some of these trials are badged as being double or triple blinded. There is a risk of unblinded clinicians transferring attitudes or providing differential treatment to active and sham arms. 16

Feasibility for use in routine practice

To our knowledge, no previous trial provided data on the human resource requirements to deliver this treatment. There is no guidance on required facilities or clinical governance, and there are no reported assessments on the impact of integrating this treatment into routine clinical care. Hence, there are still not many recommendations on how to set up and deliver this treatment.

Who should deliver low-level laser therapy?

The reported studies used physicians, dentists or physiotherapists to deliver LLLT. To our knowledge, there is no information on which health-care professionals should deliver this intervention to patients.

Lack of multicentre studies

Only one small study involving 30 participants attempted to use more than one site. 17 In this trial, 28 participants were recruited at one site and one each at the other two sites. The potential benefits of a multicentre RCT would be a larger number of participants and greater variety of locations, which may increase the generalisability of the findings. There is some evidence18 that single-site clinical trials with continuous outcomes show larger intervention effects than multicentre trials. This may reduce the generalisability of the results. 18,19

Acceptability to patients

There is little evidence on the acceptability to patients or on how they would perceive the LLLT treatment. In their systematic review, Oberoi et al. 15 recognised that this intervention requires patients to co-operate with an intervention delivered to an inflamed, painful oral cavity. Patients would undergo this repeatedly during a demanding, prolonged course of cancer treatment, with potential logistical and financial implications.

Acceptability to staff

Low-level laser therapy requires specialist equipment, a dedicated room and specially trained staff, as well as a degree of co-ordination for the appointments. There are no data on the acceptability of this treatment for clinicians or any occupational health hazards it entails.

Oncological safety

There are few published data on the safety of this treatment with regard to its effects on recurrence or persistence of disease.

Cost

Finally, there is little guidance on the cost of setting up and delivering LLLT, or its cost-effectiveness.

Primary objective

• To compare the clinical effectiveness of LLLT plus standard care with that of sham LLLT plus standard care, as measured by the Oral Mucositis Weekly Questionnaire-Head and Neck Cancer (OMWQ-HN), in adult HNC patients receiving (C)RT.

Secondary objectives

• To determine the clinical effectiveness of LLLT in preventing severe OM during radiotherapy or CRT for HNC as shown by the clinician-measured WHO Oral Mucositis Grading Scale scores.

• To apply evidence derived from the trial to inform NHS guidance in the use of LLLT for managing OM.

• To investigate the short- and long-term benefits to patients in terms of dependence on feeding tubes, nutritional status, pain control, admission to hospital, treatment interruptions, swallowing function and QoL.

• To investigate the long-term risks of LLLT (e.g. survival, recurrence and disease progression).

• To identify barriers to and facilitators of implementing LLLT in routine clinical care through a qualitative process evaluation.

Economic evaluation

• To compare the total costs of LLLT with those of sham LLLT, calculated by combining data collected from the electronic case report form (eCRF), Health Service Utilisation and Time and Travel Questionnaires (see Report Supplementary Material 1 and 2) with nationally available unit cost data.

• To compare quality-adjusted life-years (QALYs) derived from the responses to the EuroQol-5 dimensions, five-level (EQ-5D-5L), questionnaire with those of the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire 30 (EORTC QLQ-C30) measured at baseline and throughout the trial.

• To compare the cost-effectiveness measured in terms of the incremental cost per change (improvement) in OMWQ-HN score recorded between baseline and at 6 weeks of therapy (as detailed in the statistical primary end point).

• To evaluate incremental cost per QALY of LLLT when compared with standard care (from the perspective of the NHS and Personal Social Services over 14 months).

Qualitative study

• To identify barriers to and facilitators of recruitment by interviewing patients, interviewing health professionals, observing launch event and site initiation visits, and audio-recording recruitment consultations.

• To feed back to sites barriers and facilitators that have been identified by developing a detailed action plan and preparing site-specific feedback.

• To understand practitioners’ and sites’ experiences of training in and delivering LLLT and the ‘fit’ of LLLT within the treatment pathway.

• To identify barriers to and facilitators of wider implementation of trial findings and LLLT.

Inclusion criteria

• Adults aged ≥ 18 years diagnosed with HNC.

• Patients who had the capacity to provide written informed consent.

• Patients who had received a histological diagnosis of squamous cell carcinoma of the oral cavity, oropharynx, nasopharynx, larynx, hypopharynx or unknown squamous cell primary of head and neck origin histologically confirmed.

• Patients who had been discussed in a head and neck MDT meeting and were deemed medically fit for an agreed treatment plan for primary or adjuvant radiotherapy ± concurrent or induction chemotherapy (cisplatin or cetuximab).

• It had been planned for the patient to receive a minimum of 60 Gy to a defined clinical target volume in the oral cavity or oropharynx, or neck levels Ia/b, as defined by the current Radiation Therapy Oncology Group criteria. 20

Exclusion criteria

• Patients who were known to be pregnant or planning to become pregnant within the trial treatment period.

• Patients who had photosensitive epilepsy.

• Patients who had parotid tumours.

• Patients who had previous radiotherapy for HNC.

• Patients who were experiencing current/ongoing OM and trismus, limiting laser access for treatment.

• Patients who are experiencing active heavy tumour bleeding from their mouth (haemorrhage).

• Patients for whom the MDT recommend short-course palliative radiotherapy.

• Patients on immune suppressant drugs (except low-dose steroids).

• Patients who were participating in other trials assessing different treatments for OM.

• Patients who were unable to provide written informed consent.

Low-level laser therapy

Participants were scheduled to receive three sessions of LLLT (or sham) per week for 6 weeks, with each session taking 20–30 minutes. The LLLT sessions took place prior to radiotherapy, ideally no longer than 2 hours before (in protocol version 2.0 this was 60 minutes and this was amended to 2 hours for protocol version 2.1) and a minimum of 24 hours after the last session. The LLLT was delivered via a non-contact method that involved shining a weak laser light on areas inside the oral cavity. Throughout the treatment, the participant had to be in a reclined position while keeping their mouth open.

At each session, 20–30 prespecified spots were treated with LLLT. Spots were located within the following anatomical sites of the oral cavity: hard and soft palate (four spots), ventral tongue and floor of the mouth (four spots), buccal mucosa (six spots), labial mucosa (four spots), dorsal tongue (six spots) and lateral border (six spots). The treatment of each spot required the laser to be shone on it for 60 seconds. The primary tumour site was avoided and a minimum of 20 spots could be expected to be treated at each session. Each participant had an individualised treatment exclusion diagram with any areas that must not be treated clearly marked by their treating clinician. These diagrams represented those patients who had not undergone surgery to remove the primary tumour prior to irradiation (approximately 60% of patients).

Low-level laser therapy equipment

The laser machine included a control unit (Figure 4) with an attached probe that was used to deliver the laser therapy. The probe was similar in size to a toothbrush (Figure 5) and was fitted with a new transparent sleeve at the start of each session for infection control purposes.

FIGURE 4.

The THOR^® LX2.3 660-nm dental laser and control unit (reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication).

Reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication.

FIGURE 5.

The THOR 660-nm, visible-red, single-laser dental probe (reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication).

Reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication.

Low-level laser therapy was delivered using a red laser with the following specifications: wavelength 660 nm, power output 75 mW, beam area 1.5 cm², irradiance 50 mW/cm², exposure time 60 seconds and fluence 3 J/cm². In addition to the laser system, the manufacturer supplied an accompanying sham adaptation switch box (see Laser device allocation concealment).

Training for staff delivering low-level laser therapy

Low-level laser therapy was delivered by a variety of clinical staff who had undergone the appropriate laser training, as documented on site delegation and training logs. A comprehensive training package was developed for staff to complete prior to delivering any LLLT. This included an online eLaser Training Course (provided by NHS Healthcare for Education England, www.e-lfh.org.uk/home/; accessed 16 February 2019) and practical training with the laser system. The practical training was initially provided by the machine manufacturer, but staff in the sites were able to provide this training in-house for new staff once they became more familiar with the equipment. The manufacturer was able to provide further face-to-face practical training on how to use the LLLT machine where requested.

Laser safety and maintenance

The low-level laser used in the trial is indicated for use for OM. This laser is classified as a 3b laser, which means that it does not cut or burn but may be hazardous for eye exposure. Laser operators, observers and participants were required to wear laser safety glasses while the laser was in operation, and these could be worn over prescription glasses when required. The use of laser safety glasses was documented at each session. LLLT was delivered in a locked room with reflective surfaces covered and a warning was placed outside the door.

All sites were required to appoint a laser protection advisor (LPA) if one was not already in place. The LPA role included approving the local laser rules and providing overarching advice on the general use of the laser. Sites were also required to appoint a laser protection supervisor (LPS), who had day-to-day knowledge of the laser therapy administered for the LiTEFORM trial. Their role included maintaining a list of laser operators and their training, and authorising them as competent to use the laser. Typically, this was a member of staff who was involved in the trial and named on the delegation log. The manufacturers serviced the equipment annually at no additional cost and provided a replacement laser machine at sites wherever possible during this time to allow for continued provision of service. All sites were also provided with an equipment decontamination guide for cleaning and disinfection in addition to following any local procedures.

Funding of the trial intervention

The THOR laser system [THOR LX2.3 laser and light-emitting diode (LED) therapy system (THOR Photomedicine Ltd, Chesham, UK)] was purchased by each site and was classed as an excess treatment cost (ETC). Each laser system included:

• THOR LX2.3 laser and LED therapy system – control unit

• THOR 660-nm, visible-red, single-laser dental probe

• glass laser light guide(s) – spares provided free of charge

• patient laser safety glasses

• staff laser safety glasses

• sham adaptation box

• leads/cables/connectors.

The sham adaptation switch box was purchased upfront by sites in addition to the laser system package. All sites were reimbursed for the sham switch box as a research cost through the site agreement with the sponsor. All servicing for the trial performed by THOR was free of charge. Staff training and time to deliver the active LLLT were also classed as ETCs. Costs for delivering the sham LLLT therapy sessions were classed as research costs and were included in the per-participant payment.

Standard care for both arms

Aside from the randomised allocation to receive either LLLT or sham treatment, both arms received standard of care as per local policies. This varied between sites because of the lack of a standardised protocol for the treatment of OM in the UK (see Appendix 1, Table 29, for site-specific practices). However, this typically included patient education through reinforcing the importance of good oral hygiene and hydration, as well as providing nutritional advice and pain management using analgesics, mouthwashes and coating gels.

The variation in practice across the country regarding the prevention and treatment of OM was acknowledged during the design of the LiTEFORM trial, and no attempt to standardise care across sites was made. This was to ensure that sites were comfortable recruiting participants to the trial, knowing that they were going to be given all standard measures at that site in addition to LLLT/sham treatment. Given that the evidence for treatment of OM is, to our knowledge, very limited, this seemed to be reasonable.

Primary outcome measure

The primary outcome measure was the results of the OMWQ-HN. 21 The OMWQ-HN is a validated patient-reported outcome measure (PROM) with proven sensitivity and responsiveness in comparison with other patient-reported measures. 22 It has been recommended by the Head and Neck Steering Committee, part of the Coordinating Center for Clinical Trials, National Cancer Institute, Rockville, MD, USA. 23 There is good evidence of high completion rates, with patients returning over 90% of questionnaires, even during the last 2 weeks of radiotherapy when patients are at their lowest ebb. 22

The OMWQ-HN results were collected at baseline (after consent but before the first day of LLLT treatment), weekly during radiotherapy and at 4 and 14 months post radiotherapy. The OMWQ-HN is a nine-item patient-reported questionnaire specific to the HNC population, and measures symptoms of mucositis, including mouth and throat soreness, and their impact on patient well-being over the past 7 days. All questions use a Likert-type response format. The first question quantifies the mouth and throat symptoms that the patient is experiencing on a five-point scale, with 0 indicating no soreness and 4 indicating extreme soreness. There then follow five questions addressing the impact of soreness on patient function (sleeping, swallowing, drinking, eating and talking), which are rated on a five-point scale, with 0 indicating no limitations and 4 indicating unable to do. The remaining three questions assess the degree of mouth and throat pain and soreness using an 11-point scale, with 0 indicating no pain or soreness and 10 indicating the worst pain or soreness imaginable or possible. The responses to the OMWQ-HN are summed to give a total overall score between 0 and 54 points, with a higher score indicating poorer well-being and oral function. The decision to use a PROM for the primary outcome was guided by comments from the review board, as opposed to the original plan of a clinician-rated mucositis score. The patient and public involvement (PPI) group members were instrumental in deciding which PROM should be used; they were given a selection of PROMS that had been vetted for reliability, validity and responsiveness. The OMWQ-HN was selected by the PPI group because it was quick and tapped into meaningful things for patients at this time. The PPI group ruled out other PROMs for a variety of reasons. For example, one had a poor English translation and was felt to be ambiguous, onerous and more related to broader QoL issues that were not appropriate for people going for daily radiotherapy.

Secondary outcome measures

Economic and qualitative outcomes are detailed in Chapter 4, Economic evaluation, and Chapter 5, Qualitative study.

The WHO’s Oral Mucositis Grading Scale score24 is a clinician-rated score that measures objective, subjective and functional aspects of OM based on clinical observations, an oral examination of erythema and ulceration, and functional status. Data on the WHO scale were collected at baseline, weekly during the 6-week treatment period, and at the 4-month follow-up. The WHO scale is a single item that is scored on a five-point scale (Table 1). Owing to the subjective nature of this score, an intraoral photograph was taken at the time of the completion of the WHO mucositis score at the 4-month follow-up visit. Another member of the research team anonymised this for independent fully blinded evaluation.

The MD Anderson Dysphagia Inventory (MDADI) is a patient-reported swallowing outcome measure specifically designed for the HNC population. 25 The MDADI contains 20 items that constitute four subscales: global, emotional, functional and physical. Each item is scored on a five-point Likert scale ranging from 1 (strongly agree) to 5 (strongly disagree). Scores for one emotional item and one functional item were reverse scored in accordance with the scoring guidelines. Each subscale was calculated as the average of its items and rescaled to range from 20 (worst impairment) to 100 (no impairment). A subscale score was computed using ‘participant subscale mean’ imputation if at least half of its items were non-missing. A composite score was computed to summarise overall impairment on the emotional, functional and physical domains. The composite score was computed as the weighted average of the untransformed (before rescaling) emotional, functional and physical domain scores and was then scaled to range from 20 to 100. The composite score was computed if all three subscale scores were available. The MDADI was collected at baseline, 6 weeks and at the 4- and 14-month follow-ups.

The European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire 30 (EORTC QLQ-C30) (version 3.0) and EORTC Quality of Life Questionnaire Module for Head and Neck Cancer (EORTC QLQ-H&N35) were collected at baseline, 6 weeks and at the 4-month and 14-month follow-ups. The EORTC Quality of Life Questionnaire (QLQ), an integrated system for assessing the health-related QoL of cancer patients, comprises the following: the QLQ-C30 module, which contains 30 items that constitute 15 subscale scores – a global health status/QoL subscale, five functional subscales and nine symptom subscales (six of which are made up of single items). 26 All items are scored on a Likert scale from 1 (not at all) to 4 (very much), except for those related to global health status/QoL, which are scored from 1 (very poor) to 7 (excellent). The 15 subscales were each rescaled to range from 0 to 100. Higher scores on the global health status/QoL represent better QoL. Higher scores on the functional subscales represent higher levels of functioning. Higher scores on the symptom subscales represent greater symptomology/problems. We summarised 13 of the 15 subscales (i.e. excluding ‘global health status/QoL’ and ‘financial difficulties’) with a QLQ-C30 summary score; computation of the summary score was performed only when at least half of the subscale scores were non-missing. 27 The symptom subscales that formed part of the summary score were reverse scored to ensure that higher scores on the summary score reflected better functioning and lower symptomology.

The EORTC QLQ-H&N35 module is a diagnosis-specific module designed to be used in conjunction with the EORTC QLQ-C30. 27–29 It is intended for use among a wide range of HNC patients with disease of varying stage and being treated with various modalities. The EORTC QLQ-H&N35 contains 35 items that constitute 18 symptom subscales (11 of which are represented by single items). Higher scores on the subscales represent greater symptomology/problems.

Performance Status Scale for Head and Neck Cancer Patients (PSS-HN) is a three-item scale designed to evaluate functional performance of HNC patients on the domains of normalcy of diet, eating in public and understandability of speech. 30 The PSS-HN responses were collected at baseline, weekly during the 6-week treatment period and at the 4- and 14-month follow-ups. Each item is scored on an ordinal scale ranging from 0 to 100, with higher scores representing better functional performance. Responses on each item were dichotomised according to whether the participant had scored ≤ 50 or > 50. 30,31 For the eating in public item, a separate category was used to code participants who were inpatients.

The timed water swallow test (WST)32 was used to measure changes in swallow function. The WST provides an indication of overall swallowing performance. Participants were asked to drink 100 ml of water as they were timed and the number of swallows taken was recorded. Three measures of swallowing performance were computed: capacity (i.e. total volume swallowed divided by total time taken in seconds), volume (i.e. total volume swallowed divided by total number of swallows) and speed (i.e. total time taken divided by total number of swallows). If the participant displayed overt signs of significant aspiration or became distressed, the test was halted and the remaining amount of water in the cup was recorded. Participants scored 0 on all three WST outcomes if they had severe dysphagia or odynophagia and were nil by mouth. The WST was collected at baseline, 6 weeks and at the 4- and 14-month follow-ups.

Pain outcomes included the use of analgesics and topical treatments in mouthwashing assistance visits to an oral hygienist, as well as scores on the pain domains of the EQ-5D-5L and the OMWQ-HN.

The EQ-5D-5L ‘pain/discomfort’ item is a measure of patients’ self-reported pain/discomfort and is rated along an ordinal scale from 1 (no problems) to 5 (extreme pain or discomfort). 33

Data on the use of analgesics were categorised as no analgesia, anti-inflammatory analgesic/paracetamol (e.g. ibuprofen), opioids (e.g. morphine) and others. Data on the use of mouthwash were categorised as no mouthwash, simple mouthwash (e.g. saline), analgesic mouthwash [e.g. benzydamine hydrochloride (Difflam; Mylan UK Healthcare Ltd)], antiseptic mouthwash (e.g. chlorhexidine), mucosa-protecting mouthwash [e.g. oral mucoadhesive (Mugard, Norgine UK Ltd)] and others. Data were collected at baseline and weekly during the 6-week treatment period.

Weight and body mass index (BMI) changes from baseline were recorded on a weekly basis during treatment and at the 4- and 14-month follow-ups.

Participants’ oral intake as a proportion of normal (pre illness) and dependence on a feeding tube was recorded weekly during the 6-week treatment period and at the 4-month and 14-month follow-up visits.

Adverse events (AEs) attributed to LLLT and clinical complications included the number of days as an inpatient, the number of hospital admissions and the number of interruptions in CRT treatment.

Data on disease recurrence and persistence of disease were recorded at 14 months.

Safety

Adverse effects were recorded from day 1 of LLLT to the 12-week follow-up visit, serious adverse events (SAEs) were reported up to the last trial visit (at either 4 or 14 months) and any serious adverse reactions (SARs) were reported until trial closure. All events were graded according to severity (i.e. mild, moderate, severe or life-threatening) and their relationship to LLLT assessed (i.e. unrelated, unlikely, possible, probable or definite). Full guidance on AE and SAE reporting was provided in the protocol (see www.journalslibrary.nihr.ac.uk/programmes/hta/1557160/#/).

It was expected that most of the AEs that occurred during the trial would be related to the CRT that participants were receiving rather than the LLLT. However, it was anticipated that participants may experience the following AEs after receiving LLLT:

• nausea

• dizziness

• increase in OM symptoms within 24 hours of receiving laser therapy

• decrease in OM symptoms within 24 hours of receiving laser therapy

• tingling sensation in their mouth

• feeling of warmth in their mouth.

In the unlikely event that a participant experienced persistent or severe reaction to LLLT, staff were instructed to discontinue the intervention immediately.

Screening and recruitment

Each site held weekly HNC MDT meetings at which the MDT decided whether or not to recommend (C)RT treatment to a patient. Research team members were embedded in the MDT and identified participants who would potentially be eligible following treatment recommendations. Staff at each site screened these potential participants against the eligibility criteria. Potentially eligible participants were then approached at one of their routine appointments prior to starting (C)RT and given a participant information sheet (PIS) (see Report Supplementary Material 3) to read and consider in their own time. In addition, a video-recording of a role-play of the consent discussion between a clinician and a patient was accessible for patients at sites, as well as on the trial website (www.liteform.org.uk; accessed 23 May 2019). The PIS was also available on this website.

Sites kept logs of screening activity, including the number of patients who were screened for eligibility and given PISs.

Consent procedure

Patients were reapproached about the trial at one of their subsequent standard hospital visits prior to attending for their planned (C)RT. A minimum of 48 hours was required to have elapsed since receipt of the PISs.

Informed consent discussions were undertaken by an appropriate member of site staff (as named on the delegation log) and patients were asked for verbal consent to audio-record these discussions. Patients were encouraged to discuss the PIS with an appropriate member of site staff and were given the opportunity to ask any questions.

Those indicating that they wished to participate gave written informed consent by signing and dating the trial consent form (see Report Supplementary Material 4), which was witnessed and dated by a delegated member of the local research team. Completed consent forms and eligibility checklists were sent securely to the Newcastle Clinical Trials Unit (NCTU) to be checked for accuracy and completeness prior to randomisation. Participants who declined to take part in the trial were given the option to providing a reason and this was recorded.

Schedule of events

Figure 6 and Table 2 detail the participant flow and schedule of events, respectively. Baseline assessments were performed after consent at a standard care (C)RT planning appointment (and could be split across several appointments if needed), but always before the first day of LLLT treatment. During weeks 1–6 of the participant’s scheduled (C)RT, assessments were conducted weekly. These assessments typically took place on the same weekday as the first day of LLLT and sites were instructed to perform these assessments before LLLT was given that day. At 6 weeks, when the primary outcome was collected, further questionnaires and assessments took place.

FIGURE 6.

Trial flow chart showing the planned progress of participants.

Subsequent follow-up visits were designed to align with participants’ standard care visits. No assessments were performed at the week 12 visit, but AEs were recorded and concomitant medications were checked and recorded. At the 4- and 14-month follow-up visits, questionnaires and other assessments were repeated. Clinical outcomes regarding recurrence and disease progression were also recorded at the 14-month visit. In protocol V4.0 (amendment 3) (see Appendix 1, Table 28, and the project web page www.journalslibrary.nihr.ac.uk/programmes/hta/1557160/#/), the planned follow-up schedule was modified to reflect 14-month data being collected only for participants who commenced laser therapy prior to 6 July 2018 (see Appendix 1, Table 28).

Withdrawal of participants

Participants had the right to withdraw from the trial at any time without having to give a reason. The principal investigator could also discontinue an individual’s participation in the trial if this was considered to be in the patient’s best interest. Participants who withdrew consent for further follow-up were included in the analysis up to the date of withdrawal. Reasons for withdrawal were recorded, where available.

Laser device allocation concealment

A sham adaptation switch box was used to conceal participants’ allocation to the LLLT or sham arm. This was a small box situated between the main laser unit and the attached probe (Figure 7). The two dials outside the switch box (labelled ‘tens’ and ‘units’) were used to select a participant-specific machine number (in the range 1–99). This controlled the delivery of either sham therapy or LLLT as required. For example, for machine number 14, the operator would turn the tens dial to 1 and the units dial to 4.

FIGURE 7.

Sham adaption for laser machine (reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication).

Reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication.

Internally, the switch box contained a circuit board laid out in a grid pattern. The rows in the grid correspond to the tens dial on the switch box and the columns to the units dial. There were two possible grids that could be used within the switch box (grid 1 or 2), with each grid containing a different set of randomly assigned on/off positions (see Appendix 1, Figure 17). The switch boxes were preset to an agreed grid by the manufacturer before delivery and the sites did not have access to the internal workings of the switch box.

Following randomisation, the secure web-based system generated an e-mail to staff at the site, notifying them of the participant’s allocated machine number. Within the randomisation system, the machine number was generated in an analogous way to a kit number or bottle number used in double-blind drug trials. Machine numbers were allocated uniquely within each site. This system ensured that, in the unlikely event of a participant being unblinded, staff would not become aware of the allocation of any other participants at their site.

Blinding

The laser safety glasses supplied to staff delivering LLLT were designed by the manufacturer to block the red light coming from the probe. This prevented staff from seeing if the machine was delivering the sham output or the active LLLT. Staff were instructed to wear the laser safety glasses before switching the laser on and not to remove them until the laser was switched off. This was primarily to keep staff safe, but also to reduce the risk of any accidental unblinding.

Participants were also instructed to wear laser safety glasses that emitted a pulsing red light inside the rims of laser safety glasses (Figure 8). The pulses emitted red light in time with the pulsing light from the laser machine. If a red light is shone on the roof of the mouth, it is possible to see/experience a faint red light travelling through the hard palate when the eyes are closed. This pulsing red light was designed to help maintain the trial blinding and stop the participant from knowing if they were receiving the sham output or the active LLLT.

FIGURE 8.

Laser safety glasses (reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication).

Reproduced with permission from THOR Photomedicine Ltd, Chesham, UK; 2021, personal communication.

Additional measures were taken to protect trial blinding, including the incorporation of additional resistors in the head of the sham LLLT probe to create warmth as if it was delivering the LLLT.

Trial Management Group

The NCTU managed the trial on behalf of the trial sponsor. NCTU responsibilities included trial set-up, obtaining regulatory approvals, facilitating and performing site training, monitoring (on and off site), amendments, regular contact with site teams, maintenance of the central trial master file and trial close down.

The TMG was responsible for overseeing the day-to-day management of the trial and comprised the chief investigator, co-investigators, statisticians, health economists, qualitative researchers, a patient representative and the NCTU trial management team [i.e. trial manager, senior trial manager, clinical trial administrator and data(base) manager]. The TMG met approximately every month throughout the trial to ensure adherence to the trial protocol and monitor the conduct and progress of the trial.

Oversight committees

A Trial Steering Committee (TSC) was established to provide oversight of the trial on behalf of the funder. The TSC consisted of an independent clinical chairperson, an independent clinician, an independent statistician and a layperson. The TSC met four times throughout the trial and members were in regular contact through e-mail and teleconference when required.

An independent Data Monitoring Committee (DMC) was formed with the purpose of monitoring efficacy and safety end points. The DMC consisted of an independent clinical chairperson, two further independent clinicians and an independent statistician. DMC meetings were scheduled to take place prior to TSC meetings and the DMC made recommendations to the TSC regarding the continuation of the trial.

Sample size

The sample size calculation assumed a group mean difference of four points in the OMWQ-HN, reflecting a meaningful treatment effect, and at 6 weeks a standard deviation (SD) of the OMWQ-HN of 10.7 points. 20 The trial was powered with a 5% alpha and 90% power. The original sample size calculation required 152 participants with primary outcome data in each treatment arm. This was inflated to 190 patients recruited in each arm (380 in total) to account for a maximum of 20% loss to follow-up or missing data.

Internal nine-month pilot phase

The LiTEFORM trial included a 9-month pilot phase with up to seven sites planned to open to recruitment during this period. A further three sites had been planned to open during the full RCT (taking the total number of sites to 10). Meetings of the DMC and TSC were held at the end of the pilot phase to review recruitment, any barriers, participant safety and data collection to date. These committees made recommendations as to whether or not the trial should continue and were guided by the following progression criteria specified in the protocol:

• Site set-up to be complete for four pilot sites by 4 months post funding contractual start date (month 1), including the training of a minimum of two nurses or delegated staff to deliver LLLT in each site to a competent level to ensure that there are no gaps as a result of, for example, annual leave. The second phase of an additional three sites to be set up by month 6.

• The first four pilot sites recruiting, on average, 1.5 participants per month for the first 4 months post funding.

• The first four sites recruiting at full rate, on average, two participants per month from months 5–9 post funding.

• The additional three sites recruiting, on average, 1.5 participants per month during months 3 and 4.

• The additional three sites recruiting, on average, two participants per month during months 5–9.

• Completion of the OMWQ-HN at 6 weeks in at least 80% of randomised participants.

• A minimum of 100 participants recruited and randomised by the end of the 9-month pilot phase.

Data handling

Data were entered by sites on to the MACRO (Elsevier BV, Amsterdam, The Netherlands) database and were checked throughout the recruitment period to ensure that the eCRFs were as complete and accurate as possible. There were two types of validation to ensure data integrity: manual and electronic. The following types of checks were performed: range checks, consistency checks, protocol checks and accuracy checks. All issues arising from the checks were queried with site staff. All changes to the data were documented in the audit trail, including details of who made the change, when the change was made and why the change was made, to prove data integrity.

Essential data will be retained for a period of at least 5 years following close of the trial, in line with sponsor policy and the latest European Directive on good clinical practice (GCP) (2005/28/EC). 35 Data were handled, digitalised and stored in accordance with the Data Protection Act 199836 and the Data Protection Act 2018. 37 This was in accordance with General Data Protection Regulations (GDPR). For such purposes, the sponsor will act as the data controller for this study and NCTU as the data processor.

Statistical analysis plan

A complete statistical analysis plan (SAP), which provides full details of all statistical analyses, variables and outcomes, was finalised and signed before the final database lock and analysis (see the project web page: www.journalslibrary.nihr.ac.uk/programmes/hta/1557160) (see Report Supplementary Material 5).

Owing to the trial under recruiting, the statistical analyses performed were descriptive in nature and no formal statistical testing between arms was carried out. Analyses followed the intention-to-treat (ITT) principle, including all participants randomised into the trial, regardless of their adherence to the entry criteria, subsequent discontinuation of laser therapy or deviation from the protocol. The ITT analysis set was modified (mITT) to allow the seven participants who randomly received treatment that was not what they should have received, according to the original randomisation schedule, to be included in the treatment arm corresponding to the treatment they received (see Serious breach affecting treatment delivered).

Statistical analyses were conducted on complete cases from the mITT analysis set, that is participants were included in the analyses if they attended the visit of interest and had evaluable data for the outcome measure of interest. Evaluable data are non-missing for the outcome measures listed in Outcome measurements, except for the three questionnaires MDADI, EORTC QLQ-C30 and EORTC QLQ-H&N35, for which simple imputation for missing questionnaire items in accordance with the questionnaire’s scoring manual was used (see Secondary outcome measures).

A Consolidated Standards of Reporting Trials (CONSORT) flow diagram was drafted to describe participant flow and retention throughout the trial (see Figure 10). Participants who discontinued with LLLT and participants who withdrew from the trial were presented in a line listing. Descriptive statistics were used to summarise participant follow-up, compliance with the LLLT schedule and baseline characteristics.

Primary outcome

The primary outcome was the OMWQ-HN total score at 6 weeks post start of LLLT. The OMWQ-HN total score was computed for all participants except those with missing data on more than one item, with the exception of participants who scored 0 (no soreness) on question 1 (who would not then proceed to the remaining questions and would be given a total score of 0). 21

Secondary outcomes

The descriptive summaries for secondary outcomes, as presented in Chapter 3, Results, primarily focused on assessments undertaken at baseline, 6 weeks and 4 months, with further descriptive summaries included in Appendix 2.

For the WHO Oral Mucositis Grading Scale score, the number and percentage of participants with each grade were presented and the difference in proportion of participants reporting grades III or IV, indicating severe or life-threatening mucositis, was reported with associated 95% CIs.

Health-related QoL questionnaires (i.e. MDADI, EORTC QLQ-C30 and EORTC QLQ-H&N35) were scored according to their manuals and any missing data were handled as recommended. Outcomes were summarised descriptively as frequencies (and percentages) or means/medians [and SDs/interquartile ranges (IQRs)]. The difference between treatment arm means at 6 weeks and treatment arm means at 4 months for these measures were reported and presented graphically (see Figure 12).

Adverse events were coded using the Medical Dictionary for Regulatory Activities terminology (MedDRA^®) and the preferred term was used for reporting. The number of AEs per participant and worst grade per participant were summarised descriptively by allocated treatment arm.

The number of participants reporting each AE was tabulated (for AEs occurring in at least 5% of participants in either treatment arm). SAEs were presented as a line listing that included the allocated treatment arm. All non-serious adverse reactions (AEs that were possibly, probably or definitely related to LLLT) that occurred were tabulated according to the allocated treatment arm.

Screening and eligibility

A total of 231 patients were identified by trial sites and screened for eligibility. Of these participants, 50 out of 231 (22%) were deemed ineligible to take part by local research staff. When available, the main reasons given for the ineligibility of screened patients were as follows:

• planned radiotherapy dose of < 60 Gy (n = 11)

• non-NHS patient (n = 10)

• OM and trismus prior to radiotherapy treatment (n = 7)

• non-squamous cell carcinoma (n = 5)

• participation in a competing trial (n = 5).

Of the 181 patients deemed eligible, 145 (80%) were given information about the trial and (20%) were not. When available, reasons for not having been given information about the trial were as follows:

• patient was missed (n = 9)

• there was no capacity at the site to take on new participants (n = 3)

• patient did not want to commit to the additional time/burden (n = 2)

• patient was participating in a competing trial (n = 2).

Of the 145 patients given information about the trial, 58 (40%) declined the offer to participate. The reasons given for non-participation (which were largely reported in free text) were as follows:

• patient did not want to commit to the additional time/burden for the trial (n = 22)

• patient felt that the trial was too much to consider after their cancer diagnosis (n = 6)

• patient did not want to be randomised to receive sham (n = 3)

• patient could not organise travel to accommodate the trial (n = 2)

• staff were unable to contact the patient (n = 2).

Randomisation by arm

Following consent, 87 participants were randomised: 44 were allocated to the LLLT arm and 43 were allocated to the sham arm.

Numbers analysed

Statistical analyses were conducted on complete cases from the mITT analysis set (Table 3) (see Chapter 2, Statistical analysis plan).

Withdrawals and deaths

All participants provided data at their baseline visit. Three participants in the sham arm withdrew from the trial before starting laser therapy (see Figure 10). During the 6-week treatment period, six further participants withdrew from the trial (LLLT arm, n = 4; sham arm, n = 2). The reasons and timings of withdrawals are given in Table 4. Two further participants died during the trial: one of metastatic squamous cell carcinoma (LLLT arm) and one of bilateral pulmonary thromboembolism (sham arm).

Participant follow-up visits

Participants were scheduled to return for their trial assessments at the following time points:

• baseline (after consent but before day 1 of radiotherapy)

• weeks 1–5

• 6 weeks

• 4 months after the final 6-week radiotherapy session (± 2 weeks)

• 14 months after the final 6-week radiotherapy session (± 2 weeks).

Only 28 participants (LLLT arm, n = 12; sham arm, n = 16) who started radiotherapy prior to 6 July 2018 had been in the study long enough to attend their 14-month trial visit.

Attendance at trial visits and compliance with trial visit windows

The attendance at trial visits for those participants still being followed up was high: 75 out of 78 (96%) at the 6-week visit and 70 out of 77 (91%) at the 4-month visit (Table 5). Nearly all participants (26/28, 83%) who had been in the trial long enough to have a 14-month visit scheduled attended. However, compliance with the ± 2-week trial visit windows at 4 and 14 months was poor: 28 out of 70 (40%) at 4 months and 14 out of 26 (54%) at 14 months (Figure 11). We suspect that the compliance with these trial visit windows reflects local follow-up protocols for cancer.

FIGURE 11.

Compliance with the 4- and 14-month trial visit windows. The solid vertical lines are at 4 and 14 months since the last radiotherapy session; the dashed vertical lines are ± 2 weeks around these visit times. At 4 months, 28 out of 70 (40%) participants were within the 2-week visit window and at 14 months there were 14 out of 26 (54%) participants.

Laser therapy sessions

Table 7 presents a summary and a breakdown of the number of laser therapy sessions received by treatment arms for (1) all 87 randomised participants and (2) the 71 participants included in the analysis of the primary outcome. The lowest reported effective frequency of LLLT sessions for OM in HNC is two per week. 45 We, therefore, considered receiving at least two sessions of laser therapy per week for 6 weeks to be sufficient to deliver a therapeutic dose. Only 54% (47/87) of participants received at least two sessions of laser therapy per week for 6 weeks [with a slightly higher percentage, 63% (45/71), for the participants included in the analysis of the primary outcome].

Discontinuation of treatment

Of the 87 randomised participants, 83 received at least one session of LLLT or sham laser therapy (see Table 7). Eighteen participants (LLLT arm, n = 6; sham arm, n = 12) discontinued treatment before completing their scheduled 18 sessions, but remained in the trial and continued with trial assessments (Table 8). The most frequently recorded reason for discontinuation was the participant becoming too unwell to tolerate the laser therapy sessions. Radiotherapy for HNC, as stated previously, is a very intense treatment with significant associated side effects. The reasons for withdrawal reflect a pattern of the side effects gradually building up. People who struggled with the laser treatment withdrew from the trial early, soon after starting laser treatment. As treatment continued, more participants then dropped out because of the side effects of their (C)RT.

Radiotherapy and chemotherapy received

Details of the therapy regimes for participants receiving radiotherapy and CRT are described in Tables 9 and 10. All participants receiving radiotherapy received intensity-modulated radiation therapy (IMRT).

Participants evaluable for the primary outcome

A total of 78 participants (LLLT arm, n = 40; sham arm, n = 38) remained in the trial at 6 weeks when the primary outcome was collected (see Figure 10). Of these participants, seven did not (fully) complete the OMWQ-HN and could not be included in the primary analysis (LLLT arm, n = 3; sham arm, n = 4). The primary analysis included 71 participants (LLLT arm, n = 37; sham arm, n = 34), that is 81% of the 87 randomised participants. The completion rate of the OMWQ-HN at each trial visit is presented in Appendix 2, Table 31.

Primary analysis

At 6 weeks, the OMWQ-HN total score in the LLLT arm ranged from 8 to 50 points (mean 33.2 points, SD 10.0 points) and in the sham arm ranged from 4 to 53 points (mean 27.4 points, SD 13.8 points) (Table 11 and Figure 12). The OMWQ-HN total score was, on average, 5.8 points higher (95% CI 0.1 to 11.5 points) in the LLLT arm than in the sham arm, with a higher score indicating poorer well-being and oral function. At the design stage of the trial, a minimal clinically important difference (MCID) between arm means was considered to be 4 points. The 95% CI provides the range of possible ‘true’ mean differences in the population of interest that are compatible with the trial data; it includes the prespecified MCID, but its lower limit is very close to a zero mean difference, so does not preclude the possibility of a very small mean difference between the arms. This analysis should be interpreted cautiously given the number of participants included and the role of chance variation.

FIGURE 12.

Distribution of OMWQ-HN total scores at 6 weeks by treatment arm (n = 71). a, Box plot of OMWQ-HN total scores at week 6 by treatment arm. The solid line represents the median and the dashed line represents the mean. Each black marker represents an individual participant. b, Histograms showing frequency of OMWQ-HN total scores at week 6 by treatment arm for LLLT (left) and sham (right).

The summary statistics for the OMWQ-HN total score for each week are given in Table 12 and participants’ individual scores are plotted over the course of treatment in Figure 13. Changes in the OMWQ-HN total from baseline to 6 weeks are shown in Figure 14.

FIGURE 13.

Individual change in OMWQ-HN total score across treatment arms. Participants in light blue were excluded from the primary analysis because of missing OMWQ-HN data at 6 weeks. Higher scores on the OMWQ-HN indicate poorer well-being and oral function.

FIGURE 14.

The OMWQ-HN total score at baseline and 6 weeks. Higher scores on the OMWQ-HN indicate poorer well-being and oral function. One participant in the LLLT arm is not included because of missing data at baseline.

Planned subgroup analyses

To inform future studies, summary statistics for the OMWQ-HN total score at 6 weeks by level of trial stratification factors, for the 71 participants in the primary analysis, are given in Table 13.

World Health Organization Oral Mucositis Grading Scale score

The number and percentage of participants with each WHO grade at baseline, 6 weeks and 4 months are given in Table 14. At baseline, no participants were assigned the highest grades III/IV, which indicate severe or life-threatening mucositis. At 6 weeks, data were available for 73 out of 78 participants (93.6%; two missing from the LLLT arm and three missing from the sham arm). In the LLLT arm, 19 out of 38 (50%) participants had grade III/IV mucositis, compared with 21 out of 35 (60%) participants in the sham arm. The proportion of participants with grade III/IV OM at 6 weeks was, on average, 10% (95% CI –32.7% to 12.7%) lower in the LLLT arm than in the sham arm. This 95% CI is wide and indicates that these trial data are compatible with a difference between treatment arms in the percentage of participants deemed to have grade III/IV OM at 6 weeks, in the population of interest, that ranges from 33% fewer participants in the LLLT arm than in the sham arm to 13% more.

Health-related quality-of-life measures

Figures 15 and 16 present the 95% CIs for the difference between treatment arm means at 6 weeks and 4 months for health-related quality-of-life (HRQoL) measures. In addition, these figures include the means and SDs used to calculate the mean differences, along with, for context and to facilitate interpretation, the baseline means and SDs for the participants who were included in those calculations at each time point. The 95% CIs at 6 weeks and 4 months are wide, reflecting the small number of participants with evaluable data and the large variation between participants at each time point. The 95% CIs indicate the range of possible treatment effects that are compatible with the trial data and include effects in both directions (in favour of LLLT and in favour of sham); all include the null effect and, therefore (and given that there are 22 outcomes at each of two time points), these results should be interpreted cautiously.

FIGURE 15.

The MDADI and EORTC scores: mean difference between treatment arms at 6 weeks and 4 months. All MDADI and EORTC items ask participants to use ‘the past week’ as a reference. Higher scores on the MDADI and EORTC QLQ-C30 reflect relatively ‘positive’ outcomes (e.g. better functioning or QoL). Higher scores on the EORTC QLQ-H&N35 (pain to felt ill) reflect relatively ‘negative’ outcomes (e.g. greater symptoms or problems). Higher percentages on the EORTC QLQ-H&N35 (painkillers to weight gain) reflect a greater frequency of occurrence. n indicates, for each treatment arm, the number of participants included in the calculation of the mean and SD at 6 weeks or 4 months (and the corresponding mean and SD for the same participants at baseline is reported too). Difference estimates are unadjusted for baseline.

FIGURE 16.

The EORTC scores: mean and percentage difference between treatment arms at 6 weeks and 4 months. All EORTC QLQ items ask participants to use ‘the past week’ as a reference. Higher scores on the EORTC QLQ-H&N35 (pain to felt ill) reflect relatively ‘negative’ outcomes (e.g. greater symptoms or problems). Higher percentages on the EORTC QLQ-H&N35 (painkillers to weight gain) reflect a greater frequency of occurrence. n indicates, for each treatment arm, the number of participants included in the calculation of the mean and SD at 6 weeks or 4 months (and the corresponding mean and SD for the same participants at baseline is reported too). Difference estimates are unadjusted for baseline.

Performance status scale for head and neck cancer patients

Table 14 shows the summary statistics for the PSS-HN subscales: normalcy of diet, eating in public and understandability of speech. Note that the summaries presented are for the available data at each time point and, therefore, the number of participants varies across time points.

Among the three subscales, participants were most burdened by a relatively reduced capacity to consume a normal diet. By 6 weeks, the majority of participants were, at best, able to manage only very soft food textures (87.8%; 65/74 with available data) and able to eat only with selected people, in selected places (80.6%; 54/67 with available data). On the other hand, participants had good speech intelligibility. The majority of participants were at least ‘understandable most of the time: occasional repetition necessary’ throughout the trial. A small percentage of participants had significant speech difficulties at 6 weeks (LLLT arm, 10.3%; sham arm, 14.3%), at which time face-to-face communication was necessary to make themselves understood. Descriptive statistics for the PSS-HN at each visit are presented in Appendix 2, Table 59.

Timed water swallow test

Descriptive statistics for the WST outcomes are displayed in Table 14 for both swallow volume and swallow capacity. Data for each visit are presented in Appendix 2, Tables 61 and 62. Participants scored zero on all WST outcomes if they had severe dysphagia or odynophagia and were nil by mouth. One participant at baseline, 12 participants at 6 weeks, two participants at 4 months and one participant at 14 months scored zero on all measures.

Data on the WST outcomes were available for 61 out of 78 participants (78.2%; eight missing from the LLLT arm and nine missing from the sham arm) at the 6-week visit and for 57 out of 77 participants (74%; 14 missing from the LLLT arm and six missing from the sham arm) at the 4-month visit. Reasons for missing WST data are presented in Appendix 2, Table 63. At the 6-week visit, the principal reasons for missing data were participants declining to take the WST (7/17; 41.2%), staff unavailability (3/17; 17.6%) and administrative/clerical errors (3/17; 17.6%). At the 4-month visit, most omissions were because of staff availability (8/20; 40%) and participants not attending the 4-month visit or the clinic session (8/20; 40%).

Weight and body mass index

Summary statistics for weight and BMI are reported in Table 14. Descriptively, there was a gradual decline in weight and BMI over time from baseline to the 4-month visit. Nutritional support is indicated for > 10% weight loss from baseline. By the 6-week visit, 8 out of 74 participants (10.8%) had lost > 10% of their baseline weight (LLLT arm, 6/39; sham arm, 2/35). By the 4-month visit, 29 out of 64 participants (45.3%) had lost > 10% of their baseline weight (LLLT arm, 12/31; sham arm, 17/33). Summary statistics for weight and BMI at each visit are presented in Appendix 2, Tables 64 and 65.

Oral intake and tube dependence

Data on feeding tube use were available for 73 out of 78 participants (93.6%; two missing from the LLLT arm and three missing from the sham arm) at the 6-week visit and for 67 out of 77 participants (87%; six missing from the LLLT arm and four missing from the sham arm) at the 4-month visit. Summary statistics for oral intake levels and feeding tube status at each visit are presented in Appendix 2, Tables 66 and 67.

There was a high prevalence of feeding tube dependence during radiotherapy for all participants. At the 6-week visit, there were equal proportions of participants on a feeding tube in both treatment arms [25/38 (66%) in the laser arm and 23/35 (66%) in the sham arm]. However, among those using feeding tubes at 6 weeks, the percentage of participants who were totally dependent on their feeding tube was greater in the LLLT arm (15/25, 60%) than in the sham arm (9/23, 39%) (Table 15). At the 4-month visit, 25% (17/67) of participants retained a feeding tube, with, descriptively, a higher percentage in the LLLT arm (11/33, 33%) than in the sham arm (6/34, 18%).

Pain outcomes

Analgesics use at baseline and 6 weeks is summarised in Table 16. At baseline, 67.4% of participants (58/86 with data available) reported using analgesics in the past week. At 6 weeks, 97.4% of participants (72/74 with data available) reported using analgesics in the past week. Overall, anti-inflammatory analgesics/paracetamol were the most common analgesics used, followed by opioids. Further descriptive statistics on weekly analgesics use are presented in Appendix 2, Table 68.

Oral care

Mouthwash use at baseline and 6 weeks is summarised in Table 16. At baseline, 37.2% of participants (32/86 with data available) reported using mouthwash in the past week, and at 6 weeks 89.3% of participants (67/75 with data available) reported using mouthwash in the past week. Antiseptic mouthwash was the most frequently reported form of mouthwash used at baseline. Over the course of treatment, mucosa-protecting mouthwash became the most frequently reported type of mouthwash used, followed by analgesic mouthwash, simple mouthwash and antiseptic mouthwash. Further descriptive statistics on weekly mouthwash use are presented in Appendix 2, Table 69.

Visits to an oral hygienist are summarised in Table 16. At baseline, 8.1% of participants (7/86 with available data) reported visiting an oral hygienist in the past week: all made a single visit. At 6 weeks, 6.7% of participants (5/75) reported visiting an oral hygienist in the past week: four participants made a single visit and one participant made three visits. Further descriptive statistics on visits to oral hygienist during the treatment period are presented in Appendix 2, Table 70.

Hospital admissions and outpatient visits

Hospital admissions and outpatient visits were recorded from 2 weeks to 6 weeks of the treatment period. The frequency of hospital admissions was approximately balanced between arms over this period and ranged from 5% at week 3 (4/77 participants with available data) to 19% at week 5 (14/75 with available data). Outpatient appointments were frequent and ranged from 38% of participants at week 4 (having between 1 and 10 appointments) to 49% of participants at week 2 (having between 1 and 12 appointments). Few participants reported attending the head and neck ward as an outpatient, and no participant reported any outpatient visit to an accident and emergency (A&E) department during the treatment period. Further descriptive data on hospital admissions and outpatient visits are presented in Appendix 2, Table 71.

Alcohol use and smoking

Alcohol use (e.g. wine, beer, cider and spirits) and smoking (e.g. cigarettes, e-cigarettes and roll-ups) were defined as any use over the past 7 days and were recorded from 2 weeks to 6 weeks of the treatment period. Alcohol use appeared to decline over the treatment period, with 10 out of 78 (12.8%) participants reporting use at 2 weeks, 12 out of 77 (15.6%) at 3 weeks, 6 out of 77 (7.8%) at 4 weeks, 4 out of 74 (5.4%) at 5 weeks and 4 out of 72 (5.5%) at 6 weeks. Likewise, smoking appeared to decline over the treatment period, with 9 out of 78 (11.5%) participants reporting use at 2 weeks, 7 out of 77 (9.1%) at 3 weeks, 8 out of 77 (10.4%) at 4 weeks, 6 out of 74 (8.1%) at 5 weeks and 4 out of 72 (5.5%) at 6 weeks. Further descriptions of alcohol use and smoking are presented in Appendix 2, Table 72.

Disease outcomes

At the time the trial closed, 28 participants had been in the trial for at least 14 months and were still being followed up. At the 14-month visit, data on disease outcome (disease progression, recurrence or death) were available for 26 out of 28 participants (LLLT arm, 10/12; sham arm, 16/16). For all participants, there was no evidence that disease had progressed or recurred and no participants had died.

Adverse events

In total, 28 SAEs (LLLT arm, n = 17; sham arm, n = 11) were reported across 22 participants (LLLT arm, n = 13; sham arm, n = 9) (see Appendix 2, Table 75). None of these SAEs was classed as being related to the trial intervention.

The number of AEs reported was found to be similar across allocated treatment arms (Table 17). Commonly reported AEs are tabulated by treatment arm in Table 18 and a comprehensive list is provided in Appendix 2, Table 73. The most frequently reported events were gastrointestinal disorders [including stomatitis (OM), nausea, saliva altered, dry mouth and constipation], fatigue and oral pain.

Non-serious AEs that were deemed to be possibly, probably or definitely related to LLLT are reported in Appendix 2, Table 74. We note that many of these events recorded by sites as being related to LLLT could reasonably be expected to occur in patients undergoing CRT for HNC rather than those that would be expected after receiving LLLT.

Intervention costs

Participants recruited to the trial could be allocated to one of two arms: the LLLT treatment arm or the sham treatment arm. As per protocol, both the LLLT intervention and the sham intervention had the same duration and staff component. The cost of providing the LLLT intervention was microcosted, using additional information provided by THOR48 and NHS employers ‘Agenda for Change’ pay scales (2018/19). 49 The base-case analysis assumed that the patient attended all scheduled laser sessions. Further analysis used information on session attendance recorded on the eCRFs to estimate the intervention cost for each trial participant and, therefore, to estimate the cost of delivering the intervention in the LiTEFORM trial.

Intervention costs for those randomised to receive the laser therapy session included:

• equipment required for each laser therapy session

• maintenance fee for the laser therapy equipment

• estate/facilities costs for use of a treatment room

• staff cost (per minute) for set-up and preparation for each therapy session

• staff costs (per minute) of the staff members(s) who delivered the session

• staff costs (per minute) of the staff member(s) who provided administrative support.

Calculation of the intervention costs assumed the following:

• The usual lifespan of the laser is 5 years.

• Patients receive LLLT three times per week for a period of 6 weeks.

• Each session lasts 30 minutes, with 15 minutes of set-up time.

• The equipment is serviced annually.

• LLLT is delivered to the patient by a trained radiographer (band 6 mid-point).

In addition to staff time, capital costs were calculated for the LLLT using the ‘equivalent annual cost’ methodology. 50 This method converts the initial capital cost into an annual sum that equals the resources and investment plus their opportunity cost.

The equivalent annual cost of the LLLT was calculated under the following assumptions:

• The laser is used for an average of 40 sessions per week (eight sessions per day, Monday to Friday).

• The laser is in use 52 weeks per year.

• The useful lifespan of the laser is 5 years.

• The capital costs of the laser were spread over its lifespan (5 years).

A discount factor of 3.5% was applied to account for the individual’s time preference for costs to be incurred later rather than sooner. This follows guidance for best practice. 51

Hospitalisation during the intervention period

Data on the trial participant’s use of hospital services during the intervention period (a period of 2–6 weeks), including data on both the type and the frequency of the hospital service, were collected using an eCRF.

Specifically, data were collected on the:

• number of inpatient nights

• number of outpatient visits

• number of A&E visits

• number of visits to the head and neck ward

• other hospital visits (e.g. hospital dentist visits).

Health service utilisation

A health service utilisation questionnaire was developed to capture the health-care resource use of the trial participants after the intervention period (see Report Supplementary Material 1). Participants were asked about the care that they received, retrospectively, at 4 and 14 months post randomisation. To avoid double-counting with the data on the use of hospital services during the intervention period, at the 4-month data collection point the questions specifically regarding hospital services asked the trial participants to consider hospital resource use since their last laser therapy session only. The 14-month questionnaire asked about health service utilisation in the previous 10 months only.

This questionnaire collected data on:

• inpatient and day-case admissions and length of admission

• number of outpatient visits

• number of A&E visits

• primary and community-based visits

• use of private health care/personal care

• work affected by illness.

The data collected through both the hospitalisation and the health service use questionnaires were supplemented with data collected through the eCRF.

Individual patient-level data on resource use of NHS and Personal Social Services collected via the Hospitalisation eCRF Health Service Utilisation Questionnaire (see Report Supplementary Material 1) were combined with unit costs obtained from routine data sources, such as the data collected by the Personal Social Services Research Unit (PSSRU)52 and the NHS Reference Costs 2018–19. 53 The unit costs are presented in Appendix 3, Tables 76–80. The price year for all analyses was 2019.

Medication costs

The cost of prescribed medication was also included in the trial. At each visit, participants were asked to record the name of the medication, as well as the dosage, frequency, format prescribed and start and end date (if applicable) of their prescription. Medications that were insufficiently specific to obtain costs for (e.g. ‘Vaseline’ or ‘Oxygen Therapy’) were excluded from the analysis because costs vary depending on the precise medication prescribed.

Units costs were taken from the British National Formulary and multiplied by the total units prescribed for the stated period. When there were missing or inconclusive data on any one of dosage, frequency, format prescribed, start date or end date of the medication, it was assumed that the participant had been issued one prescription for the most commonly prescribed format for the recommended time period. Information regarding prescription costs was gathered from NHS Digital’s Prescription Cost Analysis – England resources. 54 To ensure comparability with the other cost analyses, medications that were prescribed after the 4-month health services utilisation data collection point were excluded from the analysis.

Assessment of effects

Response rates

The response rates relating to participant-completed health economics questionnaires are outlined in Appendix 3, Tables 81–86. Although a progressive loss to follow-up occurred over the duration of the trial, the pattern of non-response was similar across the treatment arms.

Estimation of costs

Health-related quality of life

Sampling strategy

Our sampling strategy was informed by our current and prior experience in this area,64 our theoretical framework (i.e. NPT) and what was already known about the trial context. We aimed to achieve a balance between the spread of data (to avoid missing key events or issues) and the depth of data (a manageable data set that allows for in-depth analysis). We were also responsive to the trial context, with additional data collection in response to our emerging analysis or trial events.

Our sample of patients and staff recruited to interview was purposive, using the following criteria:

• staff – a range of those professionals involved in the LiTEFORM trial at each site, including the site principal investigator, staff involved in recruitment and those involved in delivery of the intervention, and a range of professionals, including medical, nursing, allied health professional (AHP), research nurse and other staff (Table 23).

• patients – a range of male and female patients from each site (Table 24).

The timing of both staff and patient interviews varied depending on the activity at each site; for example, a change in recruitment frequency prompted a new phase of interviews. Our sample of audio-recorded recruitment conversations was partly a convenience one because sites varied in terms of their engagement with this part of the qualitative research (Table 25). Sampling was monitored and discussed regularly at each team meeting.

Data collection: recruitment and consent

Verbal consent was obtained to audio-record the recruitment conversation with patients at the start of the discussion. All of those present gave verbal consent, including any friends and family. If anyone declined, the discussion continued without being recorded. Consent could be withdrawn at any point during the discussion. Written consent was subsequently provided within the main trial consent form and on a separate consent form for any friends and family present; if those present chose not to give their written consent, the audio file was immediately deleted (see Report Supplementary Material 4 and 9). A follow-up information sheet was provided to explain how consenting individuals could contact the research team or qualitative researcher should they change their mind about the recording (see Report Supplementary Material 10).

Qualitative data management and analysis

Interviews and recruitment discussions were, with consent, audio-recorded, transcribed verbatim and edited to ensure anonymity of the respondent. Contemporaneous field notes from non-participant observation in clinical settings were edited to ensure anonymity of the participants. Data were managed using NVivo software (QSR International, Warrington, UK). The analysis was theoretically informed by NPT and was conducted in accordance with the standard procedures of rigorous qualitative analysis,77 including open and focused coding, constant comparison, memoing,78 deviant case analysis79 and mapping. 80 A sample of data was independently coded and cross-checked: the purpose of this exercise was to identify and reflect on differences. A proportion of the data was analysed collectively in ‘data clinics’, where the research team shared and exchanged interpretations of key issues emerging from the data. The analysis was conducted by Lyndsay Lindley/Tim Rapley/Nikki Rousseau (all with a range of social science backgrounds) who were joined for some data clinics by the PPI lead Valerie Bryant (who has training in qualitative analysis).

Relationship between the process evaluation and the trial

The developing analysis was regularly discussed at TMG meetings and, when appropriate, it informed changes to trial processes. Examples include a recommendation that sites should show patients a picture of the LLLT device during the recruitment discussion (giving an indication of size and appearance) and the production of a checklist of areas that needed to be addressed during site set-up, which was provided to the post pilot phase sites.

Introducing: setting up the trial – integrating with bureaucratic and organisational orders

The time that it took from the initial contact with the trial team to the set-up of the trial at each of the sites ranged from just under 5 months to over 14 months. A range of factors affected this timeline. One participant outlined part of the process of setting up the new LiTEFORM trial service:

So, there’s lots of hoops to jump through. You have to get funding for the laser, buy the laser, then you have to get approved by the cleaning safety people. . . . You have to get medical physicists to say that it’s OK to use, that your room that you’re going to use it is OK and the people who are using it have been trained to use it. So, all those hoops you’ve got to go through.

MedCon 7

As we will see below, these ‘hoops’ also included the co-ordination work around enrolling teams and getting trust sign-off, as well as the work of finding an appropriate room and finding staff to deliver the LiTEFORM treatment.

Embedding: running the trial – integrating organisational, clinical, lay and professional orders

During the set-up phases, staff at the sites had some concerns about whether or not patients would be willing to participate, given option of the sham treatment arm:

Well I think that first of all it was an assumption on my behalf that we might struggle with the sham. Because it is such a horrible treatment, the whole idea of signing up for something where it may be a sham treatment, I wondered whether people would be willing to sign up for something. But obviously that was completely unfounded.

Other 1

As we will outline, staff across all sites found that patients who were approached about participation were generally keen to take part. Staff learnt over time to optimise the actual delivery of the LiTEFORM treatment to each patient and patients, and staff were (mostly) very positive about the treatment and its impact. Despite this, the trial did not recruit to target. Capacity to deliver the therapy was a significant challenge:

It was easy for us to recruit patients, almost all patients that we approached wanted to go into the study but we couldn’t deliver it within a very busy NHS service, in the way that we had hoped that we would able to, so it limited our recruitment.

OncCon 4

Patient flow was a central issue for all of the sites: how best to co-ordinate this, given that they had only a certain capacity to deliver the laser to a specific number of people at any given time. This issue of capacity at sites is described and explored later and was the key factor limiting recruitment to the trial. However, although capacity to deliver the LiTEFORM treatment limited recruitment, other factors made recruitment challenging.

Sustaining: post-trial futures – integrating with bureaucratic, organisational and evidential orders

For staff at the sites involved, the trial closure was ‘a real shame’, ‘sad’ and for some even ‘unfair’:

We’re very, we’re kind of, disappointed . . . We put a lot of work and effort into opening the study. It was quite, it’s an involved process. So, our charity has funded us to buy the machine. Our staff has done extensive training. We’ve rearranged how we have the departments so that we’re able to deliver it and secured rooms and carried out quite a lot in terms of infrastructure.

OncCon 7

Site staff felt that they had all worked to integrate the trial and got over the ‘big hurdle’ of setting up a new service. There was reflection on what elements could have been carried out differently, but, as noted above, given the design of the LiTEFORM trial schedule, the rate-limiting factor was seen as a lack of capacity and capability to treat any more patients simultaneously. Some noted that a separate pilot trial, rather than an internal pilot, might have managed expectations and timelines. Other than waiting to see if any potentially meaningful results could emerge, they also had a question: ‘Now the trial’s finishing, I suppose the question then is, do we just offer it as a service, full stop?’ (Den 6).

Despite the practical integration problems that they had faced, staff at many sites could see a role for LLLT going forward. Staff at most sites talked about wishing to continue to deliver LLLT after the trial as standard care. For example, one person outlined that they had ‘got a meeting arranged with our manager next week or the week after to discuss how to set up a new service’ (OncCon 8). Another staff member was going to discuss it with the team at their upcoming service development meeting and another outlined that the team was ‘very pro’. The desire to try to enable this to become standard care is as a result of staff and patient enthusiasm.

The problem that many staff faced was around how to fund this new service: how to ‘make a case’ to their trust. Some highlighted that there was, currently, ‘no tariff for this’, so implementing it was going to be difficult. However, staff at some sites hoped they could fund it, one using an ‘approximate code’. Other staff felt that, given the recent NICE statement on LLLT (which many felt was deficient),82 the recommendation could be used to support the charge of an ‘interventional tariff’. Indeed, at one site while the trial was under way, a team member had been approached by their trust, which after reviewing the guidance, asked them ‘are you providing this low-level laser and if not should we consider doing it?’ (Den6). Embedded in such a new service would be the need to regularly audit laser safety, update the training and continue a formal process for authorisation of operators, alongside developing an overarching trust policy around such optical radiation devices. However, staff at a few sites were less optimistic about the potential, noting that ‘if we’re to drive this forward we need the evidence to say that it works’ (Nurse2). Without clear evidence of benefit, especially in terms of cost-effectiveness, these staff felt that the treatment could continue to be used only within a research setting, as commissioners would not support a new service.

Whether they were optimistic or pessimistic about the potential to fund such a service, staff at all sites were aware that to sustain LLLT over time elements would have to be adapted as ‘the logistics and feasibility . . . is just quite difficult’ (Radiol11). Providing LLLT is not something that people can fit in around their daily work; instead, dedicated staff were seen as needed to deliver it. Who these staff were varied between sites, but staff at most sites felt that you could not sustain LLLT using the higher band practitioners, such as radiographers or nurses. Some staff felt that, with adequate training and support, dental hygienists or dental nurses would be ideal, given that they are ‘used to dealing with mouths’, whereas others felt that it should be carried out by health-care assistants.

Other elements would also have to be adapted, including potentially increasing the number of lasers and dedicated operators. Service staff would have to be realistic about the number of patients that they could treat at any one time, as sites found that in its current form ‘it was not sustainable to treat that many patients’ (MedCon8). In part, the issue of sustainability is tied to the amount of time it takes to treat each patient. Patients themselves asked:

If there was something that could be either made, invented or looked into to speed the whole process up, say for instance inside the mouth . . . to administer the treatment, if the laser beam could be directed to two or three places at a time it would cut down the time that you were actually in the chair.

Patient 8

Staff also discussed the idea of whether or not a more efficient way of delivering LLLT will be available in future by adapting the probe, otherwise it would not be ‘able to work in practice’ due to the limited resources with the NHS. Future work could also focus on number of sites and optimum dosage per site.

Site staff were also keen that the evidence base for LLLT still be established. They were aware that the data collected by the end of the trial would not be sufficient to make any definitive claims. They wanted to continue (given the momentum, LLLT resources available and newly gained staff expertise) some form of data collection. For example, staff at one site outlined that they would be very interested in continuing to look at the role of laser ‘even if it were in the form of a non-randomised phase two study’ (Radiol5). Given that the trial was ‘very clunky’ to deliver, staff at one site argued that the focus should shift to a multicentre cohort study comparing patients who had LLLT with those who did not. Exploring the potential of delivering such collaborative evaluations would need resources in terms of time and money. Several people asked whether or not the manufacturer, THOR, should play a role, supporting some ‘sort of phase four work’. One person outlined:

Given the amount of investment there had already been . . . it’s a pity that this bit of research isn’t going a bit further. But you could argue that the people who are manufacturing the machine and stand to make money from it, should be investing more substantially in it . . . we have disseminated this equipment into quite a number of centres who didn’t have it before we started and are not going to throw it out. I would imagine they’ll be doing something with it in a certain number of patients.

MedCon 5

Within the broader context of the NICE recommendation,82 through the trial, LLLT devices are now introduced in a number of sites across the NHS. People have seen the potential value, are engaged and are motivated to support and provide LLLT. Service provision models have been introduced and integrated in their local contexts, so LLLT is workable, albeit with adaptations in expectations of patient throughput and who delivers the treatment. For sites, there is still a clear potential need and, most importantly, willingness and opportunity to collect data to generate ‘useful information’.

Strengths and limitations

A strength of our data was the spread of data across all sites. In a trial with a reasonable number of participating sites, we obtained a comprehensive account of the difficulties of trial implementation across the sites. A further strength was the multiple staff respondents from each site (including those that delivered the LiTEFORM treatment alongside those who undertook other trial-related tasks) enabling us to document the shared perspectives on problems and tensions around delivery within each site. Patient accounts were helpful in clarifying processes at sites as well as providing insight into their perspectives of experience of the LiTEFORM treatment, and were essential to the comprehensive picture of the trial obtained. We were less successful in obtaining recordings of recruitment consultations from all sites. However, the audio-recordings obtained tended to corroborate the information from both patient and professional interviews (which suggested that in the LiTEFORM trial, formal recruitment discussions were not problematic and that patients were keen to take part in the trial). Finally, we were generally successful at obtaining follow-up interviews with both staff and patients, which enabled us to understand changes over time.

Secondary outcome measures

We compared the number of participants developing severe mucositis (WHO mucositis grades 3 and 4) in the LLLT and sham arms. The percentage of patients with a severe mucositis score was 10% (95% CI –32.7% to 12.7%) lower in the LLLT arm than in the sham arm. The CIs are too wide for us to conclude that LLLT reduced the risk of severe mucositis when compared with sham LLLT. Previous trials have shown that prophylactic LLLT significantly reduces the risk of severe OM. 22,39,40 The Laser Mucite trial41 in France recently published its results. This multicentre RCT was also too underpowered to show a difference between the sham and LLLT arms. 41

Economic evaluation

As with the statistical analysis, the economic evaluation was compromised by the small sample and lack of statistical power. However, there are aspects of the LiTEFORM trial economic component that may be useful for hospital trusts considering introducing a LLLT service for the treatment of the side effects of HNC.

The potential cost of delivering LLLT in an NHS setting has not previously been established. In the base-case analysis, the cost of delivering the intervention was estimated to be £802 per patient, based on the assumption that patients would have three 30-minute LLLT sessions per week for 6 weeks. This relatively low estimated cost of delivering the intervention was mainly because of the low yearly cost of the laser equipment (purchase price £6420), maintenance and training, with the bulk of the costs related to staff time.

The three times per week protocol chosen for the LiTEFORM trial was based on pilot work in Southampton,90 and was within Bensadoun’s guidelines13 for designing a treatment protocol. The recently published Multinational Association of Supportive Care in Cancer (MASCC) guidelines include a list of suitable protocols. 91 Although these are five times per week protocols,91 some are less time intensive than the protocol that we have chosen.

Of particular note, the Antunes et al. 39 and Oton-Leite et al. 42 treatment protocols can be completed in 11–12 minutes, rather than 30 minutes, as required for the LLLT sessions used in the LiTEFORM trial. These other protocols involve LLLT sessions on every day of irradiation rather than three times per week and, therefore, involve having to give 30 treatment sessions over 6 weeks rather than 18 treatment sessions. Although implementing a protocol such as this would mean that the number of patients who could be treated per week would increase, it is likely that the total cost of delivering the intervention would be similar to or potentially higher than the cost of delivering the protocol used in the LiTEFORM trial. This is because, although the total treatment time (810 minutes) would be approximately the same for either protocol (given an estimated 15-minute turnaround time between sessions), it is likely that more administrative time would be needed to co-ordinate the 12 extra LLLT sessions per patient around other hospital appointments and staff availability.

Qualitative study

This component of the LiTEFORM trial did recruit enough participants to fulfil its aims.

There was no lack of engagement from research teams at sites: this trial made sense to them, they wanted to be involved, and they stayed committed to it and evaluated it as worthwhile. Patients were enthusiastic about the trial and willing to participate. The qualitative work demonstrates the lengths to which sites were expected to go to to open the trial. Many sites had to apply for charitable or other funding to purchase a laser machine and then jump through many hoops to find a suitable room that was free for them to use at a range of times that conformed with laser regulations, and have it modified as necessary and install appropriate safety signage and get it approved by the local LPA. An appropriate dental chair or treatment couch had to be available. Staff who were willing to be trained in laser usage then had to be found and they had to have available time to deliver the treatment. Therefore, the trial required setting up a new type of service within already very busy departments with no additional funding.

The barriers to recruitment to this trial were logistical. In the words of one of the oncologists interviewed, ‘this was a very complex trial to do’. Capacity to deliver the therapy was the significant challenge over the life of the trial. How best to co-ordinate patient flow, given that sites only had a certain capacity in terms of the numbers of trained staff to deliver LLLT at any given time, was a central problem site staff had to manage on a day-to-day basis. They also often faced restrictions around when they could access the rooms that had been set up and approved to deliver the LLLT in. Even with a good network of communication between the trial team and the wider teams, they needed to undertake constant monitoring, adjustment and accommodation work to (re)schedule patients’ LLLT sessions alongside other treatments. Given their strong commitment to the trial, how best to manage the hidden logistical workload of this new type of service was the key issue all site teams faced.

Determinants to wider implementation of trial findings and low-level laser therapy

Only 71 participants out of the 87 recruited were included in the final analysis. Eighteen participants discontinued LLLT during the 6-week period of treatment delivery. In five cases, this appeared to be specifically related to the LLLT treatment, whereas in the other 13 cases it was because of other symptoms. Only 28 out of the 87 participants attended all 18 LLLT sessions and only 47 out of the 87 attended at least two LLLT sessions per week for the 6-week treatment period. When the themes from the qualitative study are factored in, it is clear that LLLT is not completely benign from the patient perspective. Staff talked about how they observed and reflected on patients’ ability to complete LLLT treatment once the side effects of (C)RT had become significant. Given the limited capacity to deliver LLLT, staff at some sites reported a bias to select patients for the LiTEFORM trial who they felt might be less likely to suffer severe (C)RT side effects (because they felt that they would then be more likely to complete the LiTEFORM trial protocol). If this approach was mirrored in routine NHS care, it risks limiting access to LLLT to ‘lower-risk’ patients and excluding those with greatest potential to benefit from the treatment.

Our chosen protocol for LLLT was based around published guidelines13 and pilot work at Southampton, and was similar to other protocols in the literature. 90,91 There is a paucity of data on how well tolerated LLLT is, with few data on missed LLLT sessions, so the LiTEFORM trial contributes important data in this regard. In particular, it demonstrates that LLLT sessions can themselves be challenging for patients and, therefore, LLLT protocols with quicker treatment sessions would be preferable.

The aim of developing the LLLT protocol was to provide the regimen that would place the lowest burden on patients but still be likely to have a meaningful effect. Bensadoun and Nair13 recommended that LLLT be delivered at least three times per week, with between 6 and 20 spots in the oral cavity being treated per session. MASCC updated its guidelines while the LiTEFORM trial was still open for recruitment. 91 NICE also produced guidelines during this time frame. 82 Both of these guidelines agree that there is now sufficient evidence to recommend for LLLT to be used routinely. Neither guideline goes so far as to say which particular laser characteristics should be chosen. In this regard, the MASCC guidelines91 simply suggest copying a treatment protocol from one of the trials that they list in their systematic review. In the light of the fact that this trial is non-definitive, it would be prudent to suggest using a laser therapy protocol from the list in the MASCC guidelines.

Serious breach

As noted, the trial benefited from close team working between researchers, clinicians and industry professionals. The allocation concealment procedures were borne out of those partnerships, and provided the methodological advantage of reduced risk of staff and participants becoming unblinded and knowledge of trial arm having an impact on outcome assessment. However, this came at the cost of an added layer of complexity (machine numbers allocated to determine treatment) and subsequent risk. In most RCTs, the unblinded members of the statistics team would have complete control over the procedures that protect allocation concealment. In this case, the coding device attached to the laser machines had been developed by the industry partner. The disparity in the matching of codes to treatment allocation meant that there was an interruption in recruitment at all sites as machines were all tested. It also dealt a blow to the confidence of the extended LiTEFORM trial team in sites. The early detection of this breach does highlight the diligence in the oversight of the trial.

Trial setting versus real-world setting

It is worth considering the extent to which the LiTEFORM trial was truly pragmatic. What is evident in this trial is the layer of additional work that the new service brings, in which clinicians are required to incorporate additional duties into their pre-existing job plan. In a routine NHS setting, this would be unlikely. It is more likely that new posts would be created specifically to provide this therapy. This artificial situation seen in the LiTEFORM trial is a product of the model of research in which excess treatment costs94 must be met by the trusts involved and are not provided by the funder so that they can be met from the trial budget. Trusts were not funded to recruit clinicians to provide LLLT, pay for laser devices, set aside a specific room or equip them for the treatment sessions. Trusts typically used charitable funds to buy the equipment and asked existing staff to find time to deliver the treatment. Room availability was also an issue. This observation is consistent with previous reports of ETCs being a barrier to site set-up and recruitment. 95,96

These significant constraints led to the most significant weakness of the LiTEFORM trial: its recruitment. It is clear from the data reported here, and those of other trials, that the LiTEFORM trial had set itself an unprecedented recruitment rate target. For comparison, the Laser Mucite trial41 took seven years to recruit 99 participants at seven sites in France, which were already providing LLLT treatment. In a little over 2 years, the LiTEFORM trial set up seven new LLLT services and recruited 87 participants.

Generalisability

Given the multicentre design, the LiTEFORM trial’s qualitative and economic data are generalisable across the NHS for those trusts considering setting up LLLT services for treating OM in HNC irradiation. It may also inform those wishing to set up LLLT services for OM caused by chemotherapy for other cancers and bone marrow transplantation.