Daily Time of Radiation Treatment Is Associated with Subsequent Oral Mucositis Severity during Radiotherapy in Head and Neck Cancer Patients

Mucositis, damage to the mucosa of the digestive tract, is a common toxicity of antineoplastic radiation and drug therapies, and has a significant impact on health, quality of life, and treatment cost (1). This complication is especially severe in patients with head and neck cancer receiving radiotherapy, which has been reported as the single most troublesome acute complication and has impacted nearly everyone (2–4). Very few treatment options are available, with limited efficacy and/or supporting evidence (2). There is a clear need for novel approaches to prevent and manage oral mucositis.

Radiation-induced mucositis is typically seen as an “outside-in” process, in which radiation nonspecifically targets the rapidly proliferating cells of the basal epithelium, causing the loss of ability to self-renew and ulceration (1). Although this process is multifactorial, the involved acute damage response to radiotherapy appears to be regulated by the circadian rhythm (5). For example, circadian variation of cell-cycle phases, an important determinant of radiosensitivity, has been well documented in mammalian proliferating tissues (6, 7), particularly in human oral mucosa (7). Previous animal studies (8, 9) have shown that sensitivity to radiation-induced genotoxic stress is tied to the treatment time of day, which in turn is correlated with the functional status of core circadian regulators (9, 10).

Two randomized clinical trials have been conducted on treatment time and oral mucositis in patients with head and neck cancer. Both reported about 10% lower rates of severe oral mucositis in patients treated in the morning (before 11 am) versus late afternoon (after 3 pm; refs. 11, 12). However, the differences were not statistically significant and only two broad time ranges were evaluated. These clinical trials selected time ranges based on an assumption of a monotonic rise in risk of mucositis throughout the day; however, this assumption may not necessarily hold true as many circadian regulated phenotypes vary in a cyclic pattern during 24 hours (7, 13–15). In this study, we investigated radiotherapy treatment times and oral mucositis severity in patients with head and neck cancer whose radiation treatments were categorized as seven time intervals throughout a standard treatment day (8:30 am to 4:30 pm).

The study cohort included 190 patients with histologically confirmed head and neck squamous cell carcinoma (HNSCC), who completed at least one course of radiotherapy, with and without concurrent chemotherapy, at Roswell Park Comprehensive Cancer Center (Buffalo, NY) between 2015 and 2017. We used the first treatment course only for patients who had more than one series of radiotherapy. We restricted the study to HNSCC to ensure similar dosage of chemotherapy and radiation.

Analytic data were extracted and combined from prospective databases that included daily treatment time of radiotherapy, radiation dosage, weekly self-reported assessment of oral mucositis, patient characteristics, lifestyles, cancer diagnosis, and treatment history (details are provided in the Supplementary Texts and Supplementary Fig. S1). Our analytic cohort consists of 92% of the patients with HNSCC (n = 207), and the only exclusions were due to a small percentage of missing oral mucositis (n = 2), noncurative intent of therapy (n = 5), or early treatment termination (n = 10) for reasons other than oral mucositis.

We created two datasets. The first dataset included 1,156 data records of all repeated oral mucositis measurements. The second dataset kept the earliest, most severe mucositis record for each patient (n = 190).

A typical course of curative intent radiotherapy lasted up to 35 days and was administered 5 days per week for 7 weeks. Patients receive definitive radiation with an intensity-modulated radiotherapy (IMRT) regimen (70 Gy in 35 fractions to the primary tumor and 56 Gy to the elective nodes in 35 fractions). Radiotherapy planning and delivery detail has been thoroughly described in a previous publication (16). For patients treated by radiotherapy with concurrent chemotherapy, cisplatin 100 mg/m² was intravenously infused once every 3 weeks, or cisplatin 40 mg/m² once per week. The first dose began with day 1 of radiotherapy.

The time of radiation treatment delivery (hour, minute) was recorded daily for each patient. The average treatment time was derived from day 1 until the day before each mucositis survey. The cumulative radiation dose prior to each mucositis survey was derived by multiplying the daily dose by the number of days of treatment.

We assessed patients' oral mucositis symptoms prospectively at a weekly base using a validated patient-reported Oral Mucositis Daily Questionnaire (OMDQ; Supplementary Texts; ref. 17), which included questions assessing patients' mouth and throat soreness and its impact on daily functioning. OMDQ, a valid and reliable measurement of oral mucositis severity (17), was developed as a mucositis-specific questionnaire to assess patient-reported outcomes and was based on a series of focus groups and one-on-one interviews with patients with cancer (18). This questionnaire overcomes underestimation of oral mucositis in patients whose primary cancers were located at larynx or hypopharynx, where the mucositis was difficult to observe clinically. The primary endpoint of this study, the severity of mouth throat soreness (MTS), was based on the second question of OMDQ, “During the last 24 hours, how much mouth and throat soreness did you have?” with a Likert-type scale for response, “0-no, 1-a little, 2-moderate, 3-quite a lot, and 4-extreme.” This question had a better discriminative validity in comparison to other OMDQ questions (17).

Standard therapy for oral mucositis pain at our institute consisted of baking soda/salt oral rinses, diphenhydramine lidocaine antacid solution (19), nonsteroidal anti-inflammatory medications and acetaminophen, followed by narcotics as needed.

Descriptive analyses were conducted for patient characteristics and potential confounding factors, including age at radiotherapy, sex, unhealthy lifestyles (tobacco smoking, alcohol consumption), disease site, treatment factors (dosage, duration, types), and treatment history (previous surgery, chemotherapy, radiotherapy). To determine which variables to be adjusted, we first incorporated all variables mentioned above into a generalized linear model (GLM), with the maximum MTS of each patient (n = 190) as the endpoint. Before including treatment time, the following variables had a P value ≤0.05 and were then adjusted in the final multivariable model: smoking status at diagnosis (never, former, current; as ordinal), type of radiotherapy (postoperative vs. others), sex, and treatment week (as categorical). For repeated MTS measurements, cumulative radiation dosage was also statistically significant and included in the final model.

Treatment time was categorized by 1- or 1.5-hour intervals (1 = 8:30–<9:30 am; 2 = 9:30–<10:30 am; 3 = 10:30–<11:30 am; 4 = 11:30 am–<12:30 pm; 5 = 12:30–<2:00 pm; 6 = 2:00–<3:00 pm; 7 = 3:00–<4:30 pm), ensuring adequate patients in each group for statistical evaluation, in the meanwhile capturing time slots throughout a standard treatment day. Specifically, we subdivided the times from 8:30 am to 3 pm to have about 30 patients per treatment groups of 1 to 6. Group 7 had only 14 patients, but these patients were treated at a time slot of special interest (late afternoon); therefore, they were not lumped together with group 6.

To check the stability of daily treatment time for each patient, we calculated the interquartile range and the 10th to 90th percentile range of daily treatment time across the entire treatment course. For patients with the 10th to 90th percentile range for more than 3 hours, we further checked their treatment time distribution in histogram.

We examined the association between the average treatment timing and repeated MTS measures using a linear mixed model. Treatment time was considered as categorical, which did not assume any pattern of relationship with MTS. We calculated the least square means (LSmeans) and SE for the MTS at each time category, on average and by survey week, which is a marginal average score adjusting for other factors in the model.

While using the repeated measures, MTS data provided us more statistical power; comparing patients' maximum MTS is more interpretable. This is a typical mucositis measurement used in clinical research (11, 12). We assessed the association between average treatment timing and the maximum MTS using GLM. The LSmeans of the maximum MTS for each time category was estimated by setting treatment time as categorical. A statistical testing for the significance of a linear trend between treatment time and the maximum MTS was conducted by setting the treatment time as ordinal (1–7). A sensitivity analyses were conducted by excluding patients treated after 3 pm, the small group of patients (n = 14) whose LSmeans fell out of this linear pattern. Another sensitivity analysis was conducted by excluding 57 patients with existing oral mucositis at baseline, and 1 patient with surrounding tissue dose = 120 (outlier).

The average age of the study population was 61.8 years at the time of radiotherapy, with the majority of patients being non-Hispanic Caucasian (89.7%) and males (77.9%). They were mainly (64.7%) diagnosed at clinical stage IV, and all patients (100%) were fully active at the beginning of radiotherapy. Only a few patients received prior treatment, including surgery (17.9%), chemotherapy (6.7%), radiotherapy (4.5%), or bone marrow transplantation (0%). The majority of patients were treated by radiation with concurrent chemotherapy (84.2%; Table 1).

Patient treatment times were very consistent across visits. On the basis the interquartile range, 90% of patients received their treatment within 2 hours of variation (Table 2). On the basis of the 10th to 90th percentile range of daily treatment time, 65.8% patients received their treatment within 2 hours of variation (Table 2). Among the 46 patients with the 10th to 90th range ≥3 hours, the majority of their daily treatment times were centered nicely within a 1- to 2-hour range, and the large variations were mainly due to a few visits (Supplementary Fig. S2).

Among 1,156 weekly mucositis survey records, 23.3% reported severe oral mucositis (MTS rated 3 or 4), whereas at the patient level (n = 190), 62.6% reported at least one severe oral mucositis event (maximum MTS rated 3 or 4; Table 2). This finding is consistent with the fact that each patient generally had a baseline MTS starting from 0 or 1 and tended to progress towards a more severe MTS over the course of treatment.

Using the repeated-measures MTS data, we found a statistically significant association between treatment time and MTS (P = 0.02). Treatment at 8:30 to <9:30 am was associated with the lowest MTS (LSmeans = 1.33; SE = 0.11), whereas treatment at 2:00 to <3:00 pm with the highest MTS [LSmeans (SE) = 1.83 ± 0.11; ref. Table 3]. Because the LSmeans of repeated MTS is an overall average between week 1 and week 7, the scores were lower than the maximum MTS (next paragraph). Estimates of the LSmeans by treatment week are presented in Supplementary Table S1.

Using the maximum-MTS data, we found a consistent pattern with repeated-MTS, that patients treated in the early morning (8:30–<9:30 am) had the lowest maximum-MTS [LSmeans (SE) = 2.24 ± 0.15]. The maximum-MTS increased among patients treated at later times of day, peaking in the early afternoon [LSmeans (SE) = 2.69 ± 0.16, 2.71 ± 0.17, and 2.66 ± 0.17, respectively for patients treated 11:30 am–<12:30 pm, 12:30–<2:00 pm, and 2:00–<3:00 pm; Table 4; Fig. 1]. In the small group of patients (n = 14) treated after 3 pm, the maximum-MTS [LSmeans (SE) = 2.36 ± 0.23] was equivalent to that of patients treated during 9:30 to <10:30 am. The statistical test assuming a linear trend showed a significant association between maximum-MTS and treatment timing (P_trend = 0.036). The association strengthened (P_trend = 0.01) after excluding the 14 patients treated later than 3 pm. Another sensitivity analysis done by excluding 5 patients with two modes of daily treatment time distribution both in the morning and afternoon resulted in a more significant association (P_trend = 0.027; Supplementary Table S2). The sensitivity analyses by excluding 30% patients with existing oral mucositis at baseline resulted in a consistent pattern with Table 3, although the association was not statistically significant (Supplementary Table S3). Examining the incidence of severe oral mucositis (maximum-MTS ≥3) using the raw data, we observed that 50% of patients treated between 8:30 and <9:30 am reported severe mucositis compared with 68% and 72% of patients treated between 12:30 to <2 pm and 2 to <3 pm (Table 4). When we looked at patients treated before 9:30 am or after 3 pm, which corresponded with the timing adopted in a previous RCT (11), the severe mucositis rates were 50% and 57.1%, respectively, consistent with that RCT report of 53% and 62%, respectively (11).

In addition to treatment time, the maximum MTS was also significantly associated with treatment week and type of radiotherapy (Table 4). Specifically, the mucositis score increased with the treatment week (P < 0.0001). However, the most severe mucositis did not always occur during the last week of treatment, which could possibly be the result of the pain medicine usage to alleviate oral mucositis. Patients who had postoperative radiotherapy had lower MTS than patients treated for the purpose of cure (P < 0.0001). We observed that female patients had higher maximum MTS as compared with males, and current smokers had higher mean maximum-MTS than former smokers, while never smokers had the lowest maximum-MTS, although the associations were not significant after adjusting for treatment time (P = 0.09). We did not observe significant associations (P values >0.1) between MTS and other covariates, including age, race, alcohol consumption, previous treatment, stage, site of primary cancer, concurrent chemotherapy, and total and cumulative radiation dosage.

These data provide important evidence in support of the association of oral mucositis and daily radiation treatment time in patients with head and neck cancer. A major strength and novel aspect of our study was the broad range of treatment time intervals evaluated (8:30 am–4:30 pm), allowing us to capture key time frames to find significant difference of mucositis severity as a function of treatment time of a day. We observed a peak oral mucositis severity for patients treated in the early afternoon that has never been reported. The severity of mucositis increased monotonically with the time of treatment until about 3 pm. Further studies with a larger sample size are needed to assess this association among patients treated after 3 pm, where a decrease in mucositis severity was observed in our data.

Two randomized clinical trials have compared the occurrence of severe oral mucositis rates between patients with head and neck cancer treated in the morning versus late afternoon (8–10 am vs. 4–6 pm; 8–11 am vs. 3–6 pm). Both studies assumed that the most severe oral mucositis would occur in patients treated after 3 pm and found about 10% fewer cases of severe oral mucositis in the morning group, but without statistical significance (11, 12). Consistent with this report, our data showed that patients treated after 3 pm had only moderate mucositis severity, equivalent to patients treated in the late morning.

Although further studies are warranted to verify such a pattern, theoretically it is possible that the sensitivity to radiotherapy over 24 hours might follow a cyclical pattern similar to a cosine curve, which is a typical pattern of phenotypes regulated under the circadian clock (7, 13–15). Daytime treatment hours (about 9 hours) may overlap part of this temporal curve. It should be noted that the late afternoon group in our study was represented by a small number of patients (n = 14) and therefore further research with larger sample sizes is warranted to verify our observation. However, if this pattern observed in our data holds true, the differences in severe oral mucositis rates by treatment time reported in the trials were likely underestimated due to missing the time windows that are associated with the most severe cases of oral mucositis.

Another strength of this study is that patient treatment time was very consistent throughout the course of treatment, which overcame a typical challenge of similar observational studies on repeated treatment times over many weeks (20). This facilitated the categorization of treatment hours into seven groupings and increased the statistical power. Furthermore, using prospectively collected repeated oral mucositis measurements, we conducted two complementary analyses; using repeated MTS data gave us more statistical power, whereas the maximum-MTS data allowed us to generate straightforward comparisons between patients. The results were consistent across these two modeling approaches. Although this is not an RCT, we adjusted for covariates comprehensively, thereby minimizing confounding factors. Finally, nearly all (92%) of the patients with HNSCC undergoing radiotherapy at our institution during the study period were included in our analytic cohort, thereby reducing selection bias as well as allowing for generalization to multiple different subsites, stages, and patient characteristics.

Our study has limitations. Some studies have shown that mucositis rates can vary by primary tumor location and use of chemotherapy, possibly due to the underestimation of oral mucositis in patients whose primary cancer were located in the larynx or hypopharynx, where the mucositis was difficult for clinicians to observe (21). However, the OMDQ, a validated patient-reported oral mucositis measurement tool (17), has shown similar mucositis rates and severity regardless of primary tumor location or use of concurrent chemotherapy (21). Although the underreporting issue is addressed, there may still be some measurement errors, as patients reported OMDQ weekly instead of daily. However, oral mucositis normally progressed gradually, and the MTS for adjacent days have been shown to be highly correlated (>0.8; ref. 17); therefore, weekly assessment of MTS are likely to be sufficient to catch patients' overall mucositis patterns during treatment (22). Nonetheless, this measurement error was random and likely to be independent of treatment times, resulting in attenuated association, which implies that the true effects of the treatment time on oral mucositis severity may be even stronger than suggested by our results. Although pain management may result in MTS underestimation, studies showed MTS scores were still the highest in the patients who were taking opioid analgesics (22), suggesting that pain medicine may not change patient ranking of maximum MTS that much, but may have underestimated the true differences of maximum MTS between patients, thereby attenuating true association.

These findings are consistent with the molecular evidence that many fundamental biological processes involved in cancer treatment are under circadian control, including cell division (13, 23–25), DNA damage response, repair, and apoptosis (26). In this regard, this concept of cancer chronotherapy, or treatment conducted at an optimal time of day, has been raised to improve efficacy and/or reduce toxicity (27–29). Although the endpoint of this study is not therapeutic effect directly, these data have implications for cancer chronotherapy, as toxicity and treatment response have common underlying physiologic bases; both were affected by cell damage and repair that are under circadian control. Evidence suggests that the administration of a drug at a circadian time when it is best tolerated usually achieves the best antitumor activity (30). Prior chronotherapy studies have mainly focused on chemotherapy. Tolerability of nearly 500 medications varies by up to 5-fold according to circadian scheduling, based on evidence from experimental models and/or patients (31). Significant survival benefits were reported from chrono-modulated delivery compared with conventional chemo delivery, in male patients with colorectal cancer treated by 5-fluorouracil, leucovorin, and oxaliplatin, but not in females (32). The interpersonal differences for the circadian variation in drug pharmacokinetic profile and in cell response to treatment have made it challenging to apply chrono-chemotherapy clinically (31). Radiotherapy may be more amenable to chronotherapy given the mode of administration (daily dosage within 5 minutes and targeted tissue specifically). However, further studies are warranted to verify whether the optimal timings of radiotherapy are homogeneous for everyone. Given the broad impact of circadian rhythm across many organs (15, 33, 34), further studies are worthwhile to examine the impact of treatment time on therapeutic effect for both head and neck and other cancer sites. Because circadian variation is tissue-specific (15), the best treatment time may be different by cancer sites. Subsequent studies are needed to better characterize and further optimize radiation therapeutic strategy, and appropriately designed clinical trials are required to confirm our findings.

This report highlights a simple treatment time solution for reducing severe oral mucositis in patients with head and neck cancer without extra cost. On the basis of this study, if 100 patients were treated before 9:30 am, it is possible that 21 fewer patients would experience severe oral mucositis compared with treating in the early afternoon. This may not only improve patient health outcomes and quality of lives, but also reduce the need for opioid pain medicine and significantly reduce the patient economic burden associated with mucositis (35). Although this study is focused on oral mucositis and head and neck cancer, associations between time of radiotherapy and other treatment outcomes in other cancer sites are worthwhile to be examined, as the circadian rhythm impact is broad (15, 33, 34).

We found a significant association between radiotherapy treatment time and the severity of oral mucositis in patients with head and neck cancer, with severity increased by later treatment time and peaking in the early afternoon. Our results suggest that treating patients with head and neck cancer earlier in the day may improve treatment tolerance and reduce toxicity. Additional prospective studies are warranted, as well as research to elucidate the possible benefits of coordinating treatment administration to patients' circadian patterns.

No potential conflicts of interest were disclosed.

Conception and design: F. Gu, A.D. Hutson, N.F. Schlecht, M.P. Antoch, A. Miller, A.K. Singh

Development of methodology: F. Gu, A. Miller, M.E. Platek, A.K. Singh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): W.D. Duncan, A. Platek, M.E. Platek, A.J. Iovoli, A.K. Singh

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Gu, W.D. Duncan, Y. Feng, A.D. Hutson, N.F. Schlecht, E.A. Repasky, A. Miller, M.E. Platek, A.K. Singh

Writing, review, and/or revision of the manuscript: F. Gu, M.K. Farrugia, W.D. Duncan, A.D. Hutson, N.F. Schlecht, E.A. Repasky, M.P. Antoch, A. Miller, A. Platek, M.E. Platek, A.J. Iovoli, A.K. Singh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): W.D. Duncan, A.K. Singh

Study supervision: F. Gu, A.K. Singh

F. Gu was supported by Roswell Park Comprehensive Cancer Center and NCI (P30CA016056) and the faculty start-up package. Study data were collected and managed using REDCap electronic data capture tools hosted at Roswell Park Comprehensive Cancer Center (36, 37). We thank Alexander Ostrowski's contribution for REDCap data collection.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.