Chemotherapy and radiation therapy-induced gut microbiota perturbation is associated with acute sleep disturbance among rectal cancer patients; Short Report

SUMMARY

Cancer treatment-induced gut microbial perturbation/dysbiosis have been implicated in the pathobiology of sleep disturbance; however, evidence is scarce. Eighteen newly diagnosed rectal cancer patients (ages 52-81; 10 males) completed a sleep disturbance questionnaire and provided stool samples for 16s RNA gene sequencing during chemo-radiotherapy. Descriptive statistics, Wilcoxon test and regression analyses were computed. Regression analyses showed the Shannon’s diversity index to be a significant factor associated with sleep disturbance. This preliminary work suggests that the biological “gut-brain axis” mechanism may be associated with symptoms of sleep disturbance.

INTRODUCTION

Colorectal cancer is the fourth most frequently diagnosed cancer among men and women worldwide (Bray et al., 2018). Preoperative neoadjuvant chemotherapy and radiation therapy (CRT) is part of the standard of treatment for locally advanced rectal cancer (NCCN, 2016). Adverse side effects, particularly sleep disturbance (SD), during CRT have been reported (Berger, et al., 2010). Debilitating CRT-related SD may negatively affect physical and mental health, overall recovery, and ability to adhere to and cope with their treatment and other survivorship-related challenges (e.g., return to work).

Acute CRT-related SD remains inadequately managed in part because the etiology remains poorly understood. Some studies have suggested that perturbation of the gut microbiota, known as dysbiosis, potentially leads to comorbidities including SD among cancer survivors, presumably involving the “gut-brain axis” (Jordan, et al., 2018). However, it is unclear whether there is a relationship between gut microbiome perturbation and SD despite that work demonstrating a link between gut microbiome perturbation and SD has promising clinical implications. Therefore, the objectives of this proof of concept study were to examine: a) the temporal changes in sleep disturbance scores and in diversity of the gut microbiome over the course of CRT of adults with localized rectal cancers; b) whether Shannon’s diversity index is associated with SD; and c) whether there are differences in the relative abundance of fecal microbiota at the genus level between SD and non-SD participants at the end of CRT.

MATERIALS AND METHODS

Study population

The data from this study are part of a longitudinal study that examines if perturbation of the gut microbiota is associated with symptoms among rectal cancer patients during CRT. Data from 18 participants who completed study procedures at all three time-points (before, middle [after 12 to 16 treatments], and at the end [after 24-28 treatments] of CRT) of the parent study were used. The parent study included subjects who had a new diagnosis of rectal cancer, were scheduled to receive CRT, and were at least 18 years of age or older. Exclusion criteria included history of intestinal chronic inflammatory diseases or history of previous abdominal surgery, diagnosed psychiatric and/or sleep disorders, comorbidities associated with SD (e.g. sleep apnea), use of insomnia medications, antibiotics, prebiotics, probiotics, steroids, and/or immune-suppressants agents within one month prior to sample collection at each assessment time-point. Ethics approval from both the Southeastern Academic Medical Center and the University of Puerto Rico Medical Science Campus were obtained prior to data collection. All participants included in the study provided written informed consent (IC).

Demographics and subjective assessment of SD

After obtaining IC, participants completed three self-reported measures: demographics and clinical information (i.e. age, weight, height), the Bristol Stool Form Scale (BSFS), and the 8-item Patient-Reported Outcome Measures Information System-Sleep Disturbance (PROMIS-SD), before, during, and at the end of the CRT. The BSFS is a widely used clinical and research settings’ ordinal type scale that evaluates the shape and consistency of the stool, ranging from Type 1 (separate hard lumps [hard to pass]) to Type 7 (entirely liquid, watery, no solid pieces) (Blake, Raker, & Whelan, 2016). As to the PROMIS-SD, each item was anchored by a five-point Likert-type scale response (1 = not at all to 5 = very much). Scores range between 8 and 40 where high scores mean higher SD. The BSFS and the PROMIS-SD has well-documented psychometric properties (Blake, Raker, & Whelan, 2016; Buysse et al., 2010). A decrease of more than 3 points in the PROMIS scores between time points represents a clinically relevant change (Yost et al., 2011). To optimize the analysis, we grouped participants into 2 groups based on a 3-point change in PROMIS-SD score from before to end of CRT. The SD group included participants with a 3-point increase in PROMIS-SD score, and non-SD participantswere those with less than a 3-point change in SD score.

16s RNA gene sequencing microbiome assay

At each of the three study time points, participant collected approximately 5 g of stool using a sterile plastic container. Briefly, 16s RNA gene sequencing microbiome assay procedures included extraction of DNA from stool bacteria for 16S RNA amplicon sequencing following the Power Soil DNA Isolation kit, (MoBio, Carlsbad, CA), and amplification of the V3/V4 region of 16S bacterial rDNA using the MiSeq Illumina platform following the Illumina protocol. We obtained a total of 9,152,072 reads with an average of 169,483 reads per sample. The sequences were processed in mothur for ambiguous sequences, homopolymer repeats and the filtered sequences were classified using the 16S RDP Database-II implemented in mothur. For all calculations, 20,000 reads were randomly selected from each sample.

Statistical analysis

Descriptive statistics were performed on demographics and clinical characteristics of the sample as well as for the PROMIS-SD for the sample as a whole and by SD group. A non-parametric (Wilcoxon or Mann–Whitney) statistical test with a significance level of ≤ 0.05 was used to evaluate the temporal changes in SD scores and in diversity indexes. Regression analyses were performed to examine the relationship of Shannon’s index and SD. The SD data was analyzed using Statistics Package for Social Sciences SPSS, version 24.0 for windows. The diversity and relative abundance of the gut microbiome was assessed using the vegan package in R.

RESULTS

Participant characteristics

Demographics and clinical characteristics are presented in Table 1.

Table 1.

Demographic and Clinical Characteristics of Study Participants

Characteristics	N (%)
Gender
Male	10 (56)
Female	8 (44)
Stage
II	1 (06)
III	17 (94)
Chemotherapy
5-FU 225 mg/m2 over 24 Hours	11 (61)
Oral Capecitabine 825mg/m2 twice/day, 5 days per week	7 (39)
Racial categories
Non-Hispanic White	8 (44)
Hispanic White	7 (39)
African American	2 (11)
Asian	1 (06)
Marital status
Married/partnered	12 (67)
Single/divorced	4 (22)
Widow	2 (11)
Occupation
Working	13 (72)
Retired	2 (11)
Handicap	2 (11)
Unemployed	1 (06)
	Mean
Age	59 ± 11.34
Number of Radiotherapy days	28 ± 4.4

Sleep disturbance

The median total scores (range 8-37; higher scores equals more severity) on the PROMIS-SD for the 18 participants at each time point are presented in Table 2. Follow-up tests using Wilcoxon test showed no significant pairwise differences for the total sample on the severity of SD during treatment (W-1.5, p = 0.14), nor at the end of treatment (W-1.4, p = 0.16) compared to the median scores before the initiation of CRT. The median SD score for the SD group (n=10) did change significantly from before (median = 12) to the end of CRT (median = 30, p ≤ 0.05).

Table 2.

Medians for scores for PROMIS-SD and for alpha diversity indexes for the total sample (n=18)

Parameter	Before	Middle	End
	Median	Median	Median
PROMIS-SD	20.0	25.0	25.5
Observed OTUS genera	120	98	101
Shannon	2.55	2.33	2.38

There were no statistical differences in the SD scores (median oral capecitabine: 26 vs. median 5-FU: 26) or in stool form (median oral capecitabine: 5 vs. median 5-FU: 6) between the two chemo-radiation groups (p>0.05). While there were no associations between SD scores and stool form (p>0.05), worse SD was associated with higher body max index (r=0.51, p=0.03).

Diversity of the gut microbiome, and fecal microbial ecology of SD patients at the end of treatment

Table 2 also shows the median values for each computed alpha diversity indexes for the total sample, namely, (a) Shannon’s (accounts for both abundance and evenness of the species present), (b) and the number of observed bacterial genera at the end of treatment. No significant differences were found between those on infusion of 5-FU vs. those on oral capecitabine in any of the diversity index at the middle nor at the end of CRT, which suggests that both groups are equivalent with respect to chemotherapy (p > 0.05). Furthermore, there were no associations between alpha diversity indexes, and stool form or body max indexes (p>0.05). The regression model showed that decrease in the Shannon’s index was a significant factor associated with increased SD scores (R² = .024, p = 0.04). Further, a visual inspection of the heat plot (Figure 1) suggested that there may be differences in the relative abundance at the genera level between SD and non-SD groups at the end of CRT (Figure 2).

Figure 1.

Heat plot showing the relative abundance of bacteria at the genus level between sleep disturbance (SD) (Y) and non-SD (N) participants at the end of treatment

Figure 2.

Bacterial genera with significantly different abundance between sleep disturbance (SD) (Y) and non-SD (N) participants at the end of treatment based on the Mann–Whitney test

DISCUSSION

CRT-related SD is a debilitating and complex symptom experienced by rectal cancer patients. Our findings suggest that participants experienced worsening of SD during CRT treatment. The development of severe SD may be attributed to multiple influencing cancer-related factors (e.g. concurrent fatigue, disturbed circadian rhythms; Berger et al., 2010), but the relationship between gut microbiome perturbation and SD remains to be elucidated. In this study, we observed a reduction of the number of observed bacterial genera and diversity through CRT (suggesting dysbiosis). Shannon’s diversity was significantly associated with SD from pre-to post-treatment. However, the finding that the R² value was 0.024, indicating a low explained variance, was unexpected. Future larger studies should consider using a multi-dimensional self-report of sleep, (e.g. Pittsburgh Sleep Quality Index) in conjunction with an objective measure (e.g. Actigraphy); it is possible that our measure did not capture the part of SD that may have had a stronger association with dysbiosis. It would be useful for further study to examine/compare associations between dysbiosis and SD in larger samples, including other groups of patients affected by non-gastrointestinal malignancies (e.g., lung) undergoing CRT. Further, a causal link between dysbiosis and SD, as well as mediation factors, remains to be determined and examined in larger samples. Conversely, it is also possible that the gut microbiota is affected by SD (Parkar, Kalsbeek, & Cheeseman, 2019). Future mechanistic studies (e.g. markers of gut mucosal barrier dysfunction, immune and, inflammatory responses related to dysbiosis) may strengthen the hypothesis of a shared “gut-brain axis” in the pathobiology of SD and co-existent symptoms in rectal cancer patients.

Our taxonomic analysis at the end of CRT showed statistically significant differences in the bacteria genera profiles between SD and non-SD groups. The majority of the differences among the bacteria present in the gut of SD patients compared to non-SD was represented by the Dialister genera. Interestingly, Dialister have been found to be associated to cirrhosis-associated duodenal dysbiosis (Chen et al., 2016) and infections (e.g. human papillomavirus infection; Ritu et al., 2019). Further, a recent review of the literature found evidence in colorectal cancer (CRC) patients that enrichment of certain strains of Fusobacterium (e.g. F. nucleatum) was associated with increased levels of proinflamatory cytokines (e.g. tumor necrosis factor), tumor progression, and poorer outcomes (Shang, & Liu, 2018). Fusobacterium was also significantly abundant in our participants. This is remarkable given that alterations in proinflamatory cytokines have also been found to be associated with SD, and that inflammation and immune dysregulation are a proposed mechanism to explain SD (Khanijow et al., 2015). The majority of the differences among the bacteria present in the gut of non-SD participants compared to SD participants was represented by the Turicibacter genera. This is important because there is evidence from studies of animals fed with high-fat diets that the Turicibacter genera was positively correlated with the intestinal short-chain fatty acidbutyric acid (Zhong, Nyman, & Fåk, 2015). Butyric acid may have beneficial effects in the host, including a role in colonic anti-inflammation, and improvement in metabolic health parameters (Zhong, Nyman, & Fåk, 2015). Larger studies are needed to determine the role of particular bacteria in susceptibility to SD.

LIMITATIONS

Limitations of this this proof of concept study include the relatively moderate sample size. A larger sample size would have permitted examining other variables of interest that could ,account for variance in the diversity of gastrointestinal bacterial populations and/or the SD experience (e.g., stress, pain, age, circadian rhythms, diet, excersice, gastrointestinal disturbances [abdominal pain, mucositis], tumor characteristics [location, stage]; Jordan, et al., 2018; Paulsen et al., 2017; Krueger, & Opp, 2016) that were not included in the analysis; however, collection of data on some variables is currently ongoing; on other variables, data will be collected in future studies. Although our initial results should be interpreted with caution because of the study limitations mentioned, they do have potential clinical implications. For example, emerging data from clinical trials suggests that exercise, the administration of specific probiotics (e.g. Lactobacillus gasseri CP2305), or consumption of foods that are a rich source of fiber and polyphenols may favorably modulate the gut microbiota (e.g. maintenance of intestinal permeability) and potentially mediate beneficial effects such as improvements in SD (Parkar et al., 2019; Nishida et al., 2017; Paulsen et al., 2017 ).

CONCLUSION

Cancer and CRT both have been linked to dysbiosis and to SD. Our findings suggested significant differences in the gut bacterial composition between SD and non-SD participants. Investigating the biological “gut-brain axis” mechanism underlying the relationship between gut microbiome dysbiosis and SD is an important next step. Additionally, possible routes for correcting dysbiosis such as pharmacological and lifestyle interventions might be part of our future research.