Introduction

Radiation-induced nausea and vomiting (RINV) are common side effects of radiation therapy for gastrointestinal cancers that negatively impact quality of life (QOL) [1]. RINV necessitates costly supportive care interventions and, in severe cases, leads to treatment delays that can compromise tumor control. Despite increased recognition of the importance for supportive care in cancer and clinical antiemetic guidelines highlighting RINV as an understudied area, little progress has been made towards understanding the mechanisms underlying RINV [2].

The neurotransmitter serotonin is thought to be the most important chemical mediator of RINV [3–5]. The gastrointestinal tract houses 90 % of the body’s serotonin and is an important anatomic region with regards to RINV [6]. Radiation treatments to the upper abdomen are considered moderately emetogenic according to the ASCO and MASCC antiemetic guidelines, carrying a 60–90 % risk of inducing symptoms at some point during treatment [2, 7]. The incidence and severity of which are thought to result from a combination of radiotherapy-related factors and patient-related factors [8, 9].

Understanding of RINV stems from several observational studies. The largest observational studies of RINV to date were conducted by the Italian Group for Antiemetic Research in Radiotherapy (IGARR) that followed 1954 patients [8, 10]. Of the patients that received radiation therapy (RT) to the upper abdomen within those studies, 29 % reported vomiting and 56 % reported nausea. Other smaller observational series that followed similar patients receiving abdominal or pelvic RT reported rates of nausea ranging from 63–83 % [11, 12]. However, in these trials, patients’ experiences such as the severity and duration of individual RINV episodes were not reported.

While the gross incidence rates of nausea and vomiting during gastrointestinal radiation therapy have been captured in both observational and randomized trials, and the development of these symptoms has been shown to clearly worsen QOL [12, 13], other aspects that characterize the patient experience of these symptoms still need to be addressed. As such, patient-reported outcomes (PRO) are necessary in order to fully describe the extent to which RINV affect individuals [14]. Specifically, the duration, severity, and timing of symptoms are important factors that undoubtedly modulate the patient experience but have not been reliably captured in work to date. We performed a post hoc exploratory analysis of the timing, duration, and severity of symptoms.

Methods

Study design

A prospective study was conducted at Sunnybrook Odette Cancer Centre. The research ethics board at the center approved the study protocol, and all patients gave written informed consent.

Patients’ inclusion and exclusion criteria

All patients aged 18 years or older who were able to provide informed consent, with a Karnofsky Performance Status (KPS) of greater than 40, a histologically, cytologically, or radiologically proven gastrointestinal (GI) tumor, and were scheduled to receive neoadjuvant long-course abdominal or pelvic radiotherapy were eligible. Patients who had received prior cranial radiotherapy, or abdominal and/or pelvic RT were ineligible. Patients who were planned to receive cranial radiotherapy or additional radiation within the 7 days following on-study abdominal or pelvic radiotherapy were also ineligible.

Medical treatment

The specific radiotherapy, chemotherapy, and antiemetic treatment plans were left to the discretion of the treating oncologists. Radiotherapy was planned using computed tomography simulation.

Patient assessments

Patients were followed daily from the day of their first radiation treatment to 7 days following the completion of their scheduled treatment. For the study, nausea was defined as the feeling that one might vomit, vomiting as the bringing up of stomach contents, and retching as the attempt to bring up stomach contents without actually doing so. All episodes of nausea, vomiting, retching, and antiemetic use were recorded by patients in a diary. An individual episode was considered new only if it occurred at least 1 min following completion of the previous episode. On every treatment day, the patient met with a research assistant in person to review all episodes of RINV. They were asked to rate the severity, duration, location, and onset timing of each event. If the planned in-person meeting did not occur, attempts were made to contact the patient via telephone on the same day. QOL was assessed on a weekly basis beginning on the first day of treatment using the Functional Living Index—Emesis Quality of Life Tool (FLIE) and the EORTC QLQ-C30 (QLQ-C30) core questionnaire (Appendix 1). This provided a nausea and vomiting-specific QOL assessment (18-item FLIE) commonly employed in antiemetic research, while still capturing overall QOL across functional scales (30-item QLQ-C30). When data was incomplete, patients were prompted at the time of collection to recall the missing symptom or antiemetic data if possible. Patients were followed until the seventh day after their final treatment, or until they requested to be taken off the study.

Outcomes of interest

  1. 1)

    Nausea severity, duration, location, and onset time and vomiting onset times.

  2. 2)

    The relationship between QOL as evaluated by the FLIE and QLQ-C30 and patient-reported outcomes.

Statistical analysis

Data analysis included all patients who were followed longitudinally. The symptoms of vomiting and retching were reported as a single composite event of “vomiting” as is common within the RINV and CINV literature to enable comparisons with historical data. The incidence of patient-reported outcomes was tabulated, and the frequencies were expressed as total values and proportions of outcomes. Descriptive statistics summarized baseline and outcome data. Demographic information was summarized as mean, standard deviation (SD), median, and range for continuous variables and as proportions for categorical variables.

FLIE was collected for the past 3 days, but PRO data was collected daily. Therefore, to correlate FLIE data against the daily PRO data, we took the last 3 days of PRO records for each week to compare with the FLIE. C30 was collected for the past week, so we used full week C30 data and daily PRO data for analysis. To search for significant relationship between PRO and QOL (FLIE + C30) scores over time, general linear mixed model (GLMM) was used. The fixed effects included time (weeks) and categorical variables of PRO. Individual patient was considered as the random effect. The outcome was the time-dependent QOL values (natural log-transformation was applied as needed). p values less than 0.05 were considered statistically significant. Analyses were performed using the Statistical Analysis Software package (SAS version 9.3 for Windows).

Results

Descriptive statistics

A total of 51 patients were consented for follow-up; however, 3 patients did not receive radiation therapy. Therefore, only forty-eight patients planned to receive curative or palliative intent abdominal and/or pelvic radiotherapy alone or with concomitant chemoradiotherapy were followed longitudinally [12].

Patient demographics are listed in Table 1. All patients who received treatment were included in the analysis. Fifty-eight percent of patients received XRT alone, and 42 % received XRT and CT concurrently. Of those receiving XRT alone, 15 % received palliative and 85 % received potential curative treatment. The median number of treatment days was 22 (range 1–58). Ninety-two percent of patients radiotherapy treatments were targeted to the abdomen, and 8 % targeted the pelvis. In total, 332 episodes of nausea severity, 351 nausea durations, 322 nausea onset times, and 347 localization outcomes of nausea were documented (Table 2). In addition, 154 outcomes of vomiting onset times and 133 vomiting contents were documented for 48 patients who received radiation to the GI tract (Table 2). Out of the cumulative total of 1504 assessable days for these patients, 58 days’ worth or 3.9 % of the data was missed.

Table 1 Patient demographics, radiotherapy (RT), chemotherapy, and antiemetic details
Full size table
Table 2 Frequencies of nausea and vomiting patient-reported outcomes, including duration, severity, location, and onset time, among all available records
Full size table

The mode and median nausea severity experienced per episode was 5 (on a scale from 1 to 10). Given the variation in nausea duration data and the manner in which patients reported, post hoc categories for duration were created. The categories of “constant” and “on and off” were reported directly by patients and were defined as nausea without a clear period of respite and nausea throughout the day but with intermittent asymptomatic periods, respectively. The most common durations of nausea were 2–14 h and constant nausea. With regards to onset timing, the original intent was to capture the exact times of onset of symptoms; however, after a few weeks, patients were frustrated as it was too demanding. Patients would have to be prompted during daily meetings with research assistants in person. As such, the pragmatic decision was made to change the timing into morning (defined as onset beginning in AM hours) and evening (defined as onset beginning in PM hours). The onset time of nausea was found it be equally likely in the morning, afternoon, or all day. In contrast, vomiting was more likely to begin in the afternoon and the most common content of vomiting was food.

The location of nausea was categorized primarily by primary location of nausea and by whether or not patients listed a secondary site. The most common area patients localized nausea to was solely the abdomen. This was followed by patients who reported their nausea to be localized to the abdomen plus an additional location.

Relationship patient-reported outcomes and 18-item FLIE

The aforementioned nausea patient-reported outcomes demonstrated significant relationships with many aspects of QOL as measured by the FLIE. Increased nausea duration has a significant adverse relationship with 5 QOL items as measured by the FLIE (Q1 (p = 0.04), Q6 (p = 0.04), Q11 (p = 0.02), Q12 (p = 0.04), and Q15 (p = 0.03)) (Table 3). Patients with fewer hours of nausea duration had fewer troubles on QOL (negative coefficient for Q1 and Q12; but positive coefficient for Q6, Q11, and Q15). After adjusting for nausea duration, FLIE items/summary scores had no significant time trends, except for Q3 (increasing over time, indicating less problem). Nausea location had significant relationship with 8 FLIE items Q2, Q3, Q6, Q7, Q8, Q11, Q15, and Q16. For each above items, the most significant locations were head, head + other, abdomen + other, neck/throat/esophagus, and stomach/stomach + other (Table 3). There were significant relationship between nausea severity and FLIE item Q2, vomiting FLIE items of Q10–18, and vomiting summary score (Table 3). Patients with higher nausea severities were more likely to have more troubles on all the listed FLIE items and vomiting summary score. Vomiting patient-reported outcomes also showed significant relationships to QOL as measured by the FLIE. There were significant relationships between vomiting onset and all FLIE items and summary scores, except for Q9 and Q18 (Table 3). There were also significant relationships between vomiting episodes and all FLIE items and summary scores, where patients with more vomiting episodes were more likely to have worse QOL (Table 3).

Table 3 Relationship between FLIE (item 1–18 and Summary scores) and nausea duration, nausea location, nausea severity, vomiting onset time, and vomiting episodes over time
Full size table

Relationship patient-reported outcomes and EORTC QLQ-C30

Patient-reported outcomes demonstrated significant relationships with QOL as measured by the QLQ-C30. Nausea duration had significant relationships with C30 fatigue, pain, dyspnea, and constipation. After adjusting for nausea duration, physical, role, social functioning significantly decrease (worsen QOL) over time and emotional functioning significantly increase (improving QOL) over time. Symptom scores of fatigue, pain, dyspnea, and financial problems also increased (worsening QOL) over time; constipation and diarrhea decreased (improving QOL) over time (Table 4). Nausea location was found to have significant relationships with physical, role, and social functioning and symptoms of fatigue, appetite loss, diarrhea, and financial problems (Table 4). There were significant relationship between nausea severity and physical, role, and social functioning and symptoms of fatigue, nausea/vomiting, appetite loss, and diarrhea (Table 4). Patients with higher nausea severities were more likely to have more troubles on the above significant functioning/symptoms scores. Finally, there were significant relationships between vomiting episodes and all scores except for role and cognitive functioning, dyspnea, and insomnia (Table 4). Patients with more vomiting episodes were more likely to have worsened QOL, except for pain, constipation, and diarrhea. However, patients with more vomiting episodes were more likely to have fewer problems with pain, constipation, and diarrhea.

Table 4 Relationship between C30 scores and nausea duration, nausea location, nausea severity, vomiting onset time, and vomiting episodes over time
Full size table

Discussion

RINV research typically focuses on the incidence of symptoms and often neglects the subjective experience of events that debilitate patients. This prospective study characterizes these other aspects of nausea, vomiting, and retching patient-reported outcomes induced by radiotherapy for gastrointestinal cancers patients. Detailed nausea and vomiting incidence data for this patient cohort has been previously published [12] and will not be included in the study herein.

Antiemetic practice guidelines estimate 60–90 % patients who received upper abdominal radiation experience RINV [7]. This value is based on a number of studies, including a pair of landmark studies by the Italian Group of for Antiemetic Research in Radiotherapy and was corroborated by our incidence data [8, 10, 12]. In addition, many radiation oncologists still underestimate the risk of RINV and neglect to prescribe antiemetics according to guideline recommendations [15, 16]. While this clearly highlights the importance of RINV management, proper perspective must be taken as incidence value alone does not accurately described patients’ experiences with RINV.

In total, up to 351 episodes of nausea severity, duration, or onset time and up to 154 outcomes of vomiting onset times or contents were documented for 48 patients with a mean treatment length of 25.5 days. This means that, on any given day of treatment, up to 50 and 23 % of patients experience nausea and emesis, respectively. However, our data shows that the mean severity of nausea experienced was only 4.66 ± 2.10 (on a scale of 1–10) and that patients experience a dichotomous pattern of duration of nausea of either 0.5 h and constant nausea. This would suggest that (1) nausea severity had a large range of values at each episode and that an individual episode may not be that severe and (2) patients could have distinct episodes of nausea which were either acute (30 min in length) or unremitting. Our data also suggests that majority (51.95 %) of vomiting episodes occur in the afternoon only. This highlights the importance of taking a 5-HT3RA in the hours prior to radiotherapy [17].

Our data indicates that worse subjective experiences of RINV showed significant correlations with essentially all aspects of QOL as measured by both the QLQ-C30 and FLIE questionnaires; the more debilitating the nausea or vomiting characteristic by a patient, the more debilitation was reflected in QOL questionnaires. While this finding is not surprising, it emphasizes that characteristics other than incidence need to be considered when creating clinical antiemetic guidelines. Currently, the fact that treatments are being directed towards a heterogeneous group of primary and metastatic tumors, in addition to different anatomic sites, is not recognized in the guidelines [17]. In addition, what is missing from most trials in literature and the guidelines is the cumulative incidence and severity of RINV [18]. We consistently undervalue the importance of the subjective experience of patients such as the pattern of nausea and vomiting, whether patients are likely to vomit daily or just once during their treatment, or whether there are periods that the symptoms are more severe. These characteristics are paramount to determining how to appropriately use antiemetics and are especially important for nausea; where there is no objective all-or-none symptom such as vomiting.

The primary challenge encountered was the feasibility of this model of data collection. It was difficult for patients to tell us exact times and to know when exactly when a symptom started and stopped. Because of this, pragmatic changes had to be made with regards to how to categorize nausea and vomiting onset timing and durations as stated previously. For instance, patients who were experiencing severe nausea or vomiting were unlikely to document each episode individually. Instead, we were limited to them just reporting “constant” nausea or nausea that was “on and off”. Another challenge with this model of collection was the need for daily patient encounters. We found that without the daily encounter with patients, they would neglect to fill in their diaries on a daily basis or not fill out all of the outcomes of interest (e.g., duration, onset, and severity). However, the dilemma we faced was, although patient encountered prevented missing data, it allowed patients to simplify their diary entries knowing that a research assistant would speak to them and document their symptoms anyways. Unfortunately, this simplification prevented us from acquiring the wealth of information we hoped for.

This study is also limited by its small sample size. Given the heterogeneity of the study population, further investigation is needed to confirm our findings. Patients were treated with various radiotherapy dose fractionation schedules across several different anatomic sites. In addition, different antiemetic compounds were utilized with a wide range of doses in the study herein; the inconsistency of which prevents analysis of antiemetic efficacy. RINV symptoms past 1 week after the completion of radiation treatment were not recorded. Nonetheless, our results highlight important aspects of nausea and vomiting that plague patients receiving GI radiotherapy and suggests possible targets for future antiemetic and QOL research.

In conclusion, our study highlights important characteristics of patients’ experiences of RINV of patients undergoing radiotherapy for GI malignancies and demonstrates that patients’ worsening subjective experiences of RINV directly correlate to debilitation of QOL. Increased focus on the identification and amelioration of patients’ subjective experiences could lead to more appropriate use of antiemetic to improve patients’ QOL.