Study recruitment was performed by a clinical research associate who was independent of administration of the assessment and treatment protocol. Once consented, the clinical research associate randomized participants into an exercise or control group according to a predetermined single block, computer-generated randomization schedule originated by the study statistician. Study investigators were blinded to the randomization schedule. The speech pathologist and study participants were aware which group they were assigned. The attending head and neck surgeons, medical and radiation oncologists were blinded to the randomization assignment of their patients. All clinicians involved in the management of consented patients were aware of their participation in this trial, and the procedures as set out in the study protocol for assessment and management were followed.
Eligible participants were recruited through a tertiary head and neck cancer center in Baltimore, Maryland, USA, beginning in December 2004, closing to recruitment in May 2011 and ending data collection in August 2013. Following Institutional Review Board approval, eligible participants were identified at a weekly head and neck tumor board conference by the speech pathologist. Participants were required to be at least 21 years of age with biopsy-proven Stage III or IV squamous cell carcinoma of the oral, oropharynx, pharynx or larynx regions only to meet inclusion criteria. Only those participants who were scheduled to receive curative non-surgical treatment at the institution, and were planned for completion of combined chemotherapy and hyperfractionated radiation therapy adhering with the institution’s organ preservation protocol for Stage III-IV HNC (using traditional 3D planning techniques with concomitant cisplatin and 5-fluorouracil) were recruited in the first few years. In the final years of recruitment, due to a change within the clinic away from this protocol, only those then planned to receive the modified organ preservation protocol for Stage III-IV HNC (using IMRT inverse planning technology and delivery used image guidance with concomitant cisplatin) were recruited. Recruitment was confined to these protocols only to minimize variability from different treatment regimes.
In addition, the physicians assessed the participants’ ability to perform ordinary tasks using the Karnofsky Performance Status Scale , with a score of 50% or more required for participation. Participants were excluded if there was evidence of distant disease, previous radiation therapy, major psychiatric illness, severe dysphagia before treatment requiring parenteral nutrition, or any concurrent illness that, in the investigator’s judgment, might increase the risk associated with participation in the study.
CRT and PEG Protocol
During the study period, the center’s Organ Preservation Protocol for Stage III-IV head and neck cancer consisted of one of two curative-intent cisplatin-based chemoradiotherapy regimens (Regimen 1 and 2), due to a change in the clinics management protocols mid-way through the study. The transition from Regimen 1 to Regimen 2 occurred at the same time for both participants groups in this study and analysis confirmed no statistically significant difference in proportions of patients who received Regime 1 or 2 in either group. All patients received hyperfractionated radiation at a dose of 125 cGy delivered twice daily for 28-33 days for a total dose of 70 Gy (67-72 Gy) to the primary tumor site, 60 Gy to the entire ipsilateral and/or contralateral neck and/or supraclavicular region if they contained involved lymph nodes, and 50 Gy to an uninvolved ipsilateral neck and/or contralateral neck and supraclavicular regions. The radiation therapy used in Regimen 1 was planned using traditional 3D planning techniques, whereas Regimen 2 was planned using IMRT inverse planning technology and delivery used image guidance. All patients received a one-week treatment break after delivery of 40 Gy of radiotherapy. With chemotherapy, Regimen 1 involved concomitant cisplatin (12 mg/m2 over 1 h) and 5-fluorouracil (600 mg/m2 over 20 h), dosed daily for five consecutive days during weeks one and six of radiation therapy. Regimen 2 involved concomitant cisplatin 40 mg/m2 weekly for 6 infusions.
All participants had a PEG tube placed at least one week before treatment as per the usual standard of care in this institution. However, PEG use was not commenced until participants had difficulty maintaining adequate nutrition and hydration from an oral diet alone.
Participants’ intake and body weight were monitored weekly using their weight measured to calculate chemotherapy dosage, or more frequently if issues arose. Participants with self-reported decreased oral intake, 2–3 weeks of continued weight loss, severe odynophagia, dysgeusia and/or dysphagia initiated tube feeds in combination with ongoing oral intake based on recommendations, guidance, and instruction from the dietitian, oncology nurse, and/or speech pathologist. PEG use in this study was recorded from the time non-oral feeding commenced, not from the time of PEG insertion. All patients were encouraged to continue to eat and drink, as tolerated during CRT even after PEG use began. PEG removal occurred post-treatment when participants resumed p.o. intake with weight gain or stable weight and no PEG use over a 2–4 week period. The decision for PEG removal was made by the physician, patient and often included input from the dietitian and speech pathologist based on the above criteria.
Participants underwent a PET/CT at 8–12 weeks post-treatment. Participants with N2 or greater neck disease or positive findings on the PET/CT underwent a post-treatment neck dissection at 8–12 weeks post-treatment.
The RCT consisted of two arms including an exercise and control group. Participants in both groups were assessed across a battery of measures at baseline and 3, 6, 12, and 24 months. The assessment battery is detailed below.
The assessment protocol completed by all participants included clinician-reported outcomes and patient-reported outcomes at each assessment time point. The primary outcome variable was functional oral intake at 3 months, as assessed using the Functional Oral Intake Scale (FOIS) . FOIS ratings were made at all study time points. The SLP recorded and classified the nature of oral intake of foods (1 = normal/regular diet, 2 = soft/easy to chew foods, 3 = blenderized/pureed, 4 = non oral/PEG tube + some blenderized/pureed foods, 5 = non oral/PEG tube dependent) and fluids (1 = normal, 2 = nectar, 3 = honey thick, 4 = pudding thick, 5 = no liquids by mouth/PEG tube dependent) that were managed safely at each time point as determined by patient report, information from the patients’ medical record and/or direct observation of food and liquid intake during a clinical swallow evaluation or swallow therapy session. This functional food/fluid level information was used to determine FOIS scores. As all patients had a PEG inserted prophylactically, the FOIS score accounted for tube feeding only if in use i.e. the FOIS score reflected non-oral feeding only if this had commenced. Similarly, once a patient had ceased non-oral feeding but may have had the PEG still insitu, the FOIS score reflected their oral intake status only. Presence or absence of a feeding tube at each time point was documented, as was overall duration that the gastrostomy tube remained in place.
Oromotor function was measured using an oromotor assessment containing 69 tasks assessing strength and range of motion of the facial muscles, tongue, and palate; and included measurement of incisal opening using the TheraBite® interdental tool . Treatment toxicities, including presence and extent of mucositis, oral cavity mucosa changes, and abnormalities were rated using (a) the Oral Cavity Assessment (OCA)  where a score of 7–10 is mild dysfunction and 15–18 is severe dysfunction and, (b) the NCI Common Terminology Criteria for Adverse Events (CTCAE) . A grade of 0 indicates no symptoms and a grade of 4 indicates life-threatening consequences. Weight was tracked in pounds and pain level was recorded using responses to pain related question on the QLQ-C30 (no pain vs. pain, score 1 vs. 2–4). The European Organization for Research and Treatment of Cancer (EORTC QLQ-C30) and the Head and Neck Cancer Quality of Life module (EORTC QLQ-H&N35) were included as disease-specific, patient-reported quality of life measures.
At the baseline and 3-month time points only, all participants underwent a Modified Barium Swallow Study (MBS) with an Esophagram. MBS assessments were not conducted at later time points as the primary time point of interest was function at 3 months. Costs, additional radiation exposure, and patient burden from additional assessments led to the decision to limit MBS assessments to only the primary outcome measure time point. The MBS/Esophagram was conducted by experienced speech-language pathologists in conjunction with a radiologist. Participants were trialed on measured liquids sips, unmeasured liquid cup drinking, pudding, and cracker using Varibar® products (Bracco Diagnostics Inc., Monroe Township, NJ) as well as trial of a barium tablet (E-Z-Disk™, Bracco Diagnostics Inc. Monroe Township, NJ). Participants were studied in lateral, anterior-posterior and right-anterior-oblique positions. A number of measures were calculated from the MBS/Esophagram. Dysphagia severity was rated using the 7-point Dysphagia Outcomes Severity Scale (DOSS) where 7 = normal diet, 1 = non-oral only . The penetration–aspiration scale was calculated on all swallows with a score of 1 = normal, 2–5 = penetration, 6–8 = aspiration as assessed from the most impaired swallow across the study excluding the initial swallow . Swallow physiology was assessed using a binary scale (present/absent) across 7 oral, 12 pharyngeal, and 6 esophageal phase deficits. Timing and efficiency of the oral and pharyngeal phases of swallow were evaluated using the Oral Pharyngeal Swallow Efficiency (OPSE) measures as calculated by dividing the percent of oropharyngeal residue by the duration of combined oral and pharyngeal phase transit time with a normal OPSE efficiency score of 50 or above . MBS analyses were conducted by a number of experienced speech-language pathologists blinded to the study group. All raters underwent inter-rater reliability training beforehand by co-rating MBS studies with other raters until agreement was demonstrated.
Adherence to Treatment – Study Journal
For the exercise group only, data on therapy adherence was collected using a study journal which was completed daily during and, at least, up to 3 months post-CRT. Participants were asked to bring the study journal to each appointment. Adherence to treatment protocol was monitored by participants recording the number of days per week exercises was completed at least once, and the total number of exercise sets performed using an exercise checklist journal.
Exercise Group Protocol
Participants in the exercise group were instructed on a set of active exercises, completed twice daily, 7 days per week during CRT (exception CRT break week in week 4) and up to 3 months post-CRT. The set of exercises included oromotor strength/stretch exercises and swallow maneuvers. Also, each day participants completed the TheraBite® exercise according to the 7-7-7 protocol (7 passive range of motion stretches, performed 7 times, repeated 7 times each day) . See Table 1 for details on the swallow exercise protocol.
Throughout the treatment period, CRT patients attended a weekly swallow therapy session which included continued instruction on the exercise protocol and swallow interventions as indicated. That is, instruction on compensatory swallow strategies and diet modifications were implemented, as indicated. A strong focus is that participants were encouraged to continue to eat and drink by mouth along with PEG use. For the remaining days of each week, participants completed the exercises independently and recorded compliance. The exercise protocol took approximately 20–30 min to complete twice daily.
Control Group Protocol
The control group received no direct SLP contact during treatment and did not complete any study exercise sets. However, all participants received a TheraBite® prophylactically as per minimum standard of care set at the institution where the study was conducted. Following initial instruction on how to use the device, actual ongoing use of the device during and post-CRT was not monitored. No other prophylactic therapy was provided. If a participant in the control group was subsequently identified as having swallowing problems post-CRT by their medical care team (n = 4), they were provided with swallow therapy exercises as indicated. All were analyzed in the control group. Further information on these 4 patients is reported in results.
The sample size was determined by the statistician to be 30 participants in each group for a predicted power of 83% (2-tailed at p = .05) for a 1.0 scale point change on the FOIS between the two groups assuming a standard deviation of 1.3. This effect was selected as the smallest effect that would be important to detect, in the sense that any smaller effect would not be of clinical or substantive significance.
The analysis of this RCT was conducted in an intention to treat (ITT) approach. Clinicians entered all study findings on paper data entry forms which were scanned and saved. De-identified forms were then sent to a data entry company (Sosio Corporation, Inc., Glen Arm, Maryland) for independent double data entry verification. All statistical analyses were conducted by an independent statistician. The demographics and clinical factors at baseline were summarized overall and by each group and were compared with either Student’s t-tests or Fisher’s exact tests or Chi-square tests. The analysis of the primary variable (FOIS) and a number of the secondary variables (oromotor, incisal opening, weight, OCA, CTCAE, pain, DOSS, OPSE, FOSAD, Penetration and Aspiration scale) were computed with a generalized GEE models, by which the difference (odds ratio) of the two groups was estimated at each time points. The time to removal of PEG tube was defined from the end of CRT treatment to the time of PEG removal, censored at the date of death if the individual died with PEG on or at the date of last contact if the individual was lost to follow-up prior to the removal of the PEG tube. Kaplan–Meier estimates were used to plot the survival function and estimate the median time to PEG removal. A log-rank test was used to compare the time to removal between the two treatment groups. A p value of p ≤ 0.05 was deemed statistically significant. All analyses were done with SAS 9.4 and R 3.2.2.