Abstract
Background and Objective:
Weight loss and small intestinal bacterial overgrowth (SIBO) are common in Parkinson’s disease (PD). We aimed to study the relationship between weight loss and SIBO in PD.
Methods:
This was a cross-sectional study with a prospective, interventional component. Consecutive patients seen in the PD clinic who agreed to participate underwent extensive history, movement exam, SIBO breath testing and answered questionnaires. A subset of those in the weight loss group were treated with rifaximin for 14 days and returned 3 months later for an assessment of their weight, GI symptoms, quality of life and SIBO status. All analyses were adjusted for age and disease duration.
Results:
Fifty-one patients participated in the study; 37 without weight loss and 14 with weight loss. Total energy intake including the distribution of macronutrient intake was similar between groups while physical activity was less in those with weight loss. PD severity scores did not differ between groups; however, PD-specific quality of life scores were significantly worse for the summary index and the subscales of emotional well-being, social support and communication. The prevalence of constipation, dyspepsia and abdominal pain/discomfort was higher in those with weight loss. The prevalence of SIBO was 14% in the weight loss group and was not different between groups. Eight PD patients with weight loss were treated with rifaximin; no significant change in GI symptoms, quality of life or weight was seen 3 months later.
Conclusion:
Although a number of differences were identified in quality of life and gastrointestinal symptoms between groups with and without weight loss, SIBO was not associated with weight loss in patients with PD. Given the exploratory nature and small number of patients with weight loss, however, further study is suggested.
Keywords
INTRODUCTION
Individuals with Parkinson’s disease (PD) have lower body weights compared to age-matched healthy individuals [1]. The lower body weight may affect their functional abilities, quality of life, and increase their risk of conditions such as osteoporosis and fractures, pneumonia and pressure ulcers. Weight loss is typically related to changes in energy intake and/or expenditure. Potential contributors to weight loss in PD are multiple and include hyposmia, deranged gastrointestinal (GI) motility (dysphagia, gastroparesis, intestinal dysmotility, and constipation), bradykinesia affecting eating, difficulty chewing, increased energy expenditure, medication effects, and depression [1–5]. Despite the potential importance of weight loss on the morbidity and mortality of PD patients [6], this area remains poorly studied.
Small intestinal bacterial overgrowth (SIBO) is an increasingly recognized cause of malabsorption and weight loss and is likely an underrecognized cause of a variety of nonspecific GI symptoms resulting in diminished oral nutritional intake [7]. Disturbances in small bowel motility and gastric acid secretion are the principal predisposing factors providing a clue to patient groups at risk of this condition. Noninvasive breath tests are commonly used to diagnose SIBO. Treatment is aimed at reducing intestinal microbial density, generally with a broad-spectrum oral antibiotic, and correction of associated nutritional deficiencies when present. Recently, SIBO was found to be prevalent in PD [8]. The potential role of SIBO in weight loss occurring in patients with PD has not previously been examined.
The objective of this line of investigation is to better identify the risk factors for weight loss in PD so as to prevent or at least limit its occurrence. Our hypothesis was that SIBO contributes to the development of weight loss in patients with PD. Specific aims were to determine the prevalence of weight loss and SIBO in PD patients seen in a PD specialty clinic and to evaluate the effect of SIBO treatment in PD patients with weight loss.
METHODS
This was a cross-sectional, case-control study with an exploratory, prospective, open-label interventional component. Cases were defined as those PD patients who experienced weight loss while Controls were those PD patients who had not experienced weight loss. Weight loss was defined as ≥ 10% decrease in weight since the diagnosis of PD. Although others have suggested that a weight loss ≥ 5% over a period of 6 to 12 months signifies clinically significant weight loss [6, 9], the investigators felt that ≥ 10% since the time of PD diagnosis was a more appropriate degree of weight loss to use for this study. Regardless, none of the patients in our non-weight loss group reported ≥ 5% weight loss. While weight loss was determined from patient clinic visits using the electronic health record whenever possible, in many instances the weight loss was determined by patient self-report. All cases and controls signed written informed consent approved by the Mayo Clinic Institutional Review Board. The study was registered on clinicaltrials.gov (identifier NCT01662791).
Part 1: Cross-sectional study
The study consisted of 2 parts. In Part 1, consecutive patients seen in the PD clinic were approached to participate. After the determination of their eligibility and willingness to participate, all enrolled patients (i.e., both Case and Control groups) underwent an assessment of demographic features, weight history since diagnosis of PD and PD history. They also provided qualitative assessments of smell and taste and were administered a number of validated questionnaires including the Gastrointestinal Symptom Severity Index (GISSI) that measures the frequency, severity and bothersomeness of individual upper and lower GI symptoms over the previous 3 months and provides subscale scores for interrelated symptom clusters [10]; the Unified Parkinson’s Disease Rating Scale (UPDRS) and the modified Hoehn and Yahr (H&Y) scale to assess PD severity; the Hospital Anxiety Depression Scale (HADS) using ≥ 8 as the cut-off for identifying significant psychiatric distress [11]; the Brief Block Food Frequency Questionnaire (FFQ) to provide estimates of usual and customary dietary macronutrient and energy intake [12]; the Paffenbarger Physical Activity Questionnaire (PPAQ) [13]; and, the Parkinson’s Disease Questionnaire-39 (PDQ-39) to address aspects of functioning and well-being for those affected by PD [14]. Finally, all subjects underwent a glucose hydrogen breath test, a test used routinely in clinical practice to assess for the presence of SIBO [15]. See Supplementary Materials for additional information about the questionnaires used and the breath test methodology.
Part 2: Exploratory interventional study
Participants
Part 2 included only those individuals in the weight loss group (i.e., only the Case group). All new or existing PD patients irrespective of race, gender, ethnicity, or medication use were eligible to participate. Exclusion criteria included women of childbearing potential; those who were wheelchair-bound, had dementia or were either unable or unwilling to complete the tests; those allergic to rifaximin or treated with an antibiotic within one month of the breath test (the standard abstinence period used in our Motility laboratory and recommended by Consensus guidelines [16]; those with disorders that could be confused with SIBO based on a review of the patients medical history (e.g., irritable bowel syndrome, inflammatory bowel disease, celiac disease, gastroparesis, and chronic pancreatitis) or previous gastrointestinal surgery (except for cholecystectomy, appendectomy or herniorrhaphy); or, those with another medical condition that would preclude participation.
Intervention
These patients were offered open-label treatment with rifaximin, 550 mg PO BID for 14 days regardless of their SIBO breath test result. Thus, both normal and abnormal breath test subjects with weight loss received antibiotic treatment. The rationale for open-label treatment and treating both SIBO (+) and SIBO (–) patients relates to the preliminary nature of the study and the nonspecific nature of SIBO symptoms which tend to fluctuate over time. We speculated that if significant differences based upon response to treatment in the SIBO (+) patients compared to the SIBO (–) patients were found, this would provide important information to support a future randomized controlled trial. Rifaximin (Xifaxin®, Salix Pharmaceuticals Inc.), is an orally administered, semi-synthetic, non-systemic antibiotic [17] and is commonly used in the treatment of SIBO, although it is not United States Food and Drug Administration-approved for this indication.
Outcomes
The patients then returned for a follow-up visit 3 months after treatment with rifaximin for a final assessment of their weight, GI symptoms, PD-specific quality of life, and SIBO breath test.
Statistical analysis
In Part 1, summary scores were calculated for each questionnaire. Mean and SD was calculated for continuous variables for the stable weight and the weight loss groups. Percentage and frequency were calculated for categorical variables of the two groups. Independent t-tests were used to assess difference for continuous demographic and descriptive variables between the two groups. Chi square test was used to assess the difference of categorical variables between the two groups such as gender, race, the presence or absence of GI symptoms, the presence or absence of PD treatments, and the presence or absence of SIBO. Fisher exact test was employed when assumptions for chi-square test were not met. Because age was found to be significantly different between groups on univariate testing, multivariable linear regression models were used to assess the relationship between variables of interest between groups while adjusting for age as a potential confounder. Similarly, because the PD duration in the weight loss group and the non-weight loss (9.1 years vs. 6.3 years, respectively) bordered on statistical significance, an adjustment for disease duration as a potential confounder was also performed. A breath test was considered abnormal if a rise >20 parts per million (ppm) from baseline was observed in either H2 or CH4 [13]. We also explored a lower cut-off level of >12 ppm rise in either H2 or CH4 given previous reports using this reference standard [8].
In Part 2, the presence of GI symptoms and weight change at baseline and 3 months after treatment with rifaximin were compared by McNemar’s tests for categorical variables and paired tests for continuous variables. Wilcoxon test was used instead when assumptions for the paired t-test were not met. A p-value <0.05 was considered statistically significant.
Sample size estimate
Based on a recent report [8], we assumed a prevalence of SIBO of 50% in PD patients using a glucose hydrogen breath test cut-off value of 12 ppm. Using the same test but based on our standard cut-off value of 20 ppm, we estimated that with a prevalence of SBBO of 35% in the weight loss group and 10% in the no weight loss group (i.e., 25% difference), 49 subjects will be required in each group using a power of 80% and p-value of 0.05.
Due to slow enrollment in the weight loss group and an ending of the period of grant funding, an analysis of the existing data was performed and demonstrated no difference in SIBO prevalence between groups. As a result, it was decided to end enrollment early.
RESULTS
Patient characteristics
Patient characteristic s: No weight loss group versus weight loss group
Mean (SD).
Adjusted for age.
Adjusted for age and disease duration.
Physical activity was marginally positively skewed (1.16); therefore, median (IQR) and p-value estimated using the Wilcoxon rank-sum test are also reported: 1810 (316, 3510) vs. 1081 (300, 1372), p = 0.0723.
Factors potentially affecting weight
As shown in Table 1, total energy intake including the distribution of macronutrient intake was similar between groups. Physical activity was less, although not significantly, in those with weight loss (897.3±584.7 kcal/week vs. 2166.8±2007.2 kcal/week, respectively; p = 0.065). While a subjective assessment of both taste and smell was frequently diminished or absent in the PD patients, no difference in either was found between groups. Anxiety (6.6±4.5 vs. 4.2±2.6, p < 0.001) and depression (4.5±3.2 vs. 3.2±2.0, p = 0.004) scores were higher in the weight loss group.
The duration of PD, both from the time of diagnosis (Table 1) and the time of first symptoms (9.1±5.0 years vs. 6.3±4.1 years in the weight loss and non-weight loss groups, respectively; p = 0.076), was similar between groups as was the severity of PD as determined by the UPDRS and H&Y stage. When divided into PD severity based on a total UPDRS (I+II+III) score of ≥ 45, only one patient in the non-weight loss group and no patient in the weight loss group was considered to have severe PD symptoms. PD drug use was similar between groups including dopaminergic drug use (data not shown).
Gastrointestinal symptoms and PD-specific quality of life
Gastrointestinal symptoms: No weight loss group versus weight loss group
Mean (SD).
Adjusted for age.
Adjusted for age and disease duration.
p-values not adjusted.
PD-specific quality of life: No weight loss group versus weight loss group
Mean (SD).
Adjusted for age.
Adjusted for age and disease duration.
Small intestinal bacterial overgrowth
Abnormal SIBO breath tests: No weight loss group versus weight loss group
Weight, GI symptoms and quality of life after rifaximin treatment in the weight loss group
Quality of life and GI symptoms in the weight loss group 3 months after rifaximin treatment
Mean (SD).
At the 3 month follow-up visit, the 2 patients with abnormal breath tests were re-examined. One subject’s breath test normalized while the other remained abnormal. Both subjects continued to lose weight (0.5 kg and 0.8 kg). Of the other 6 subjects who had normal SIBO breath tests but were treated with rifaximin, two lost more weight (0.3 kg and 1.2 kg) and three gained weight (1.3 kg, 1.8 kg, and 5.9 kg). As a group, there was no significant change in weight after rifaximin treatment (p = 0.374).
DISCUSSION
Weight loss of varying degree has been described in up to 60% of PD patients as the disease progresses [2, 18–22]. In this study, 27% of the PD patients experienced weight loss. This lower prevalence likely reflects the definition of weight loss utilized but is similar to a recent report of a secondary analysis of longitudinal data from NINDS exploratory trials in PD long-term study 1 [19]. Notably, the PD patients with weight loss were significantly older than those without loss. Although weight normally declines slightly after the sixth decade in men and seventh decade in women [23], our statistical analyses were adjusted for age and significant differences between groups remained.
Weight loss has been reported to have important implications for the PD patient including reduced quality of life, functional ability and survival [6, 24]. In this study, weight loss patients had a number of higher PDQ-39 subscale scores and summary index than non-weight loss patients, indicating worse quality of life. We also found that anxiety and depression scores were higher in those with weight loss. We speculate that this difference may reflect the psychological distress associated with weight loss in the setting of PD and is unlikely to have contributed to the weight loss itself.
Many factors have been suggested to contribute to the weight loss occurring in PD including a reduction in energy intake due to the presence of GI dysmotility (e.g., dysphagia, gastroparesis, constipation), medication-related adverse effects (e.g., nausea, anorexia) and fasting associated with drug administration, difficulty self-feeding, loss of taste or smell, and depression [1, 3]. Increased energy expenditure due to PD-related dyskinesias, rigidity and tremors has also been suggested to contribute to weight loss [1, 25]. Many of these factors were evaluated in this study. Constipation, dyspepsia and abdominal pain/discomfort occurred more commonly and were of greater severity in the weight loss group. Whether these symptoms are directly responsible for the weight loss, however, will require additional investigation. As previously stated, although weight normally declines slightly with aging, our statistical analyses were adjusted for age and significant differences between groups remained. The trend towards reduced physical activity in the weight loss group may relate to the older age. The similar duration and severity of PD as determined by the UPDRS and modified H&Y stage also do not explain the weight loss. Despite altered taste and smell, no difference was found between groups. Similarly, there was no difference in total energy or individual macronutrient intake between groups. Given that neither a reduction in dietary intake nor an increase in energy expenditure was found to explain the weight loss observed, perhaps nutrient malabsorption is contributing.
The gut microbiota has been shown to regulate motor deficits and neuroinflammation in a model of PD [26] and alterations in the gut microbiota have been described in humans with PD and relates to the motor phenotype [27]. Recently, SIBO, a form of altered gut microbiota, has been suggested as a potential cause of GI symptoms and motor fluctuations in PD. We were particularly interested in the role of SIBO as a potential factor in the weight loss occurring in PD. SIBO is characterized by a variety of signs and symptoms resulting from nutrient malabsorption when ingested substances interact with the bacteria present [7]. The most important factors preventing SIBO are normal small bowel motility and gastric acid. PD is not known to be associated with a reduction in gastric acid secretion; however, most PD patients are elderly and hypochlorhydria is common among the elderly [28] as is the use of acid suppressing medications (e.g., proton pump inhibitors). In this study, while the patients in the weight loss group were significantly older, there was no difference in acid reducing medication use between groups. In contrast, GI dysfunction is an important component of the clinical manifestations of PD [3]. PD is known to affect the enteric nervous system and the autonomic nervous system [29–31], thereby providing rationale for its clinical effects on the gut. Intestinal function, motility in particular however, has been poorly studied in the PD population. The use of dopaminergic drugs in the treatment of PD can also affect the GI tract and contribute to the development of GI symptoms. In the present study, there was no significant difference in PD drug use, including dopaminergic drugs, between groups.
The diagnosis of SIBO remains in a quandary. SIBO implies a quantitative assessment of bacteria present in the small intestine. The culture of aspirated small bowel fluid has traditionally been considered the gold standard in the diagnosis of SIBO; however, several limitations of this method have hindered its acceptance in clinical practice [7]. Therefore, indirect methods of detecting SIBO have been developed as alternatives. Breath testing is the most commonly used alternative method to diagnose SIBO. Breath testing utilizes an orally ingested carbohydrate, most commonly glucose or lactulose, as a substrate. Lactulose, a non-absorbable synthetic disaccharide, has been suggested to be superior to glucose due to its ability to detect SIBO in the more distal small intestine. This benefit, however, has been disputed given concerns that the rise in hydrogen levels may simply reflect the arrival of lactulose in the cecum [32]. In the presence of excessive bacteria in the small bowel, the substrate is metabolized releasing hydrogen and other gases (e.g., methane), which are subsequently absorbed and then released into the expired air. A rise in hydrogen or methane (usually >12 or 20 ppm from baseline) within 90 to 120 minutes in the breath sample indicates SIBO [7, 15]. Several factors may also influence the results of the breath test. As such, the SIBO breath test, irrespective of substrate used has demonstrated wide variations in sensitivity and specificity, and disappointing reliability to predict the results of small bowel culture [33].
The prevalence of SIBO in PD ranges widely in the literature. Davies et al. used the lactulose hydrogen breath test to study 15 PD patients and 15 control subjects and did not find evidence suggestive of SIBO in any of the subjects; however, the criterion used to diagnose SIBO (i.e., the double peak) is uncommonly seen and has largely been abandoned [34]. Gabrielli et al. used glucose hydrogen breath testing in 48 PD patients and 36 healthy controls and, using a 12 ppm increase from baseline as the criterion for positivity, found that 54% of PD had SIBO compared to 8% of the controls [8]. SIBO was associated with PD severity and a higher prevalence of bloating and flatulence. In an attempt to enhance diagnostic sensitivity, Fasano et al. used both glucose and lactulose hydrogen breath tests in 33 PD patients and 30 healthy controls and found a prevalence of SIBO of 54% versus 20%, respectively, although the criteria for positivity were not stated [35]. SIBO was associated with an increase in unpredictable motor fluctuations. Further supporting a role of SIBO in PD was the finding of an improvement in motor fluctuations following eradication of SIBO using the antibiotic, rifaximin [35]. In a longitudinal study of 51 PD patients using the lactulose hydrogen breath test, 34 (67%) were positive initially while 42 (82%) were positive at some point during the nearly 3 years of surveillance [36]. They also performed breath testing on 15 spouses of the PD patients and found an identical 67% prevalence of SIBO. The high prevalence of SIBO seen may relate to the use of a higher dose (25 g) of lactulose than generally used (10 g). In an uncontrolled study of 103 PD patients, Tan et al. used lactulose breath testing and diagnostic criteria incorporating rises in hydrogen or methane and found a prevalence of SIBO of 25% [37]. SIBO was associated with a shorter duration of PD and a lower constipation severity score. They did not find an association between SIBO and motor fluctuations or PDQ-39 quality of life summary index scores. Most recently, Niu et al. performed glucose breath testing in 182 PD patients from China and compared the results to 200 healthy, (age-, gender-, and BMI-) matched controls [38]. They found that 30% of the PD met criteria for SIBO compared to 9% of the controls. Motor fluctuations present were higher in the PD patients with SIBO than in the patients without SIBO. We identified a SIBO prevalence of approximately 14% using a glucose breath test and cut-offs of abnormal based upon different definitions used in the published literature [39, 40]. The broad range in SIBO prevalence described likely reflects the poor standardization of the SIBO breath test and the diagnostic criteria used.
The clinical consequences of SIBO in PD are unclear. Similar to the present study, previous reports have not identified consistent differences in GI symptoms between those with and without SIBO and conflicting findings have also been found with respect to motor function in PD [8, 37]. We were particularly interested in the role of SIBO in weight loss occurring in PD but did not find a difference in the prevalence of SIBO in those with and without weight loss. Although not focusing on weight loss, the reports previously described also did not find a difference in BMI in those with and without SIBO [8, 38]. Given the similar energy intake, PD severity and generally less physical activity in our weight loss group, unless there is some other factor affecting energy expenditure, the weight loss would seem to relate to malabsorption. In the absence of a difference in SIBO occurring between the weight loss and no weight loss groups, perhaps there is another cause of malabsorption responsible [41]. Further study, including the actual demonstration of the presence of malabsorption, is needed.
Finally, 8 PD patients with weight loss were treated with a standard treatment course of rifaximin regardless of their SIBO breath test results. The subjects then returned for a follow-up visit 3 months after treatment for a final assessment of their weight, GI symptoms, PD-specific quality of life, and SIBO breath test. The rationale for open-label treatment and treating both SIBO (+) and SIBO (–) patients relates to the preliminary nature of the study and the nonspecific quality of SIBO symptoms which tend to fluctuate over time. The only significant difference we found after treatment with rifaximin was an improvement in the emotional well-being subscore of the PDQ-39; however, because of a lack of a placebo control and the small number of patients treated, this finding requires confirmation. Although a reduction in constipation severity approached statistical significance following treatment with rifaximin, the clinical significance of this finding requires additional study given the small number of subjects treated.
Strengths of this study include the inclusion of a well characterized group of PD patients, the consistent use of a rigorous definition of weight loss, the use of several definitions of SIBO based on breath testing to enhance diagnostic sensitivity, and an attempt to show a clinical response after antibiotic treatment. This study also has several limitations. We had intended to enroll twice as many patients as were actually enrolled based on an estimated 35% prevalence in SIBO in those with PD and a predicted 25% difference in SIBO prevalence in those with and without weight loss. Due to slow enrollment, an analysis was performed and demonstrated no difference in SIBO prevalence between groups. As a result, it was decided to end enrollment early. Perhaps the definition of significant weight loss that was used was too rigorous. Future studies may benefit from exploring different standards for significant weight loss. All of the other studies described that evaluated SIBO prevalence in PD also suffered from small study populations thereby limiting definitive conclusions. The weight loss in our patients was generally determined by patient self-report; however, an attempt to verify the weight loss in the electronic health record was made whenever possible. Although the accuracy of self-reported weight loss is often suspect, a study involving 550 men and women from seven geographic areas found that the individuals reported their body weights with very good accuracy [42]. The diagnosis of SIBO continues to be a limitation of all studies of this type. The optimal means of identifying SIBO, particularly for research purposes, remains to be identified. Perhaps, the use of next generation sequencing on small intestinal fluid will provide the means to more accurately identify those with clinically-relevant SIBO [43]. The ability of the Block FFQ, despite evidence demonstrating its validity for this purpose, to allow definitive conclusions is limited given the reliance upon self-report, recall and estimation of intake. Although a SIBO treatment phase was incorporated into the study in order to determine whether antibiotic treatment would result in an improvement in weight among those PD patients with weight loss, because of the small number of patients with weight loss enrolled, the results are inconclusive. Finally, because of the multiple comparisons performed in this study, it is possible that the positive associations may be due to chance. However, in this exploratory study, we reported point and variance estimates without adjusting p-values for multiple comparisons to avoid increase risk of Type II error.
In conclusion, weight loss is an important problem in patients with PD; however, the cause(s) of the weight loss remain unclear. While the present study did not find an association between weight loss and SIBO, the results are limited by the small number of patients with weight loss enrolled. Given the exploratory nature of this study, further study is suggested. Because of the difficulty identifying patients with weight loss of >10%, however, any further study of this area should involve multiple centers so that a sufficient number of patients are enrolled.
CONFLICT OF INTEREST
None of the authors has any relevant conflicts of interest to declare.
Footnotes
ACKNOWLEDGMENTS
We would like to acknowledge the assistance of Amy Duffy, CCRP and Michaele Menghini, CCRP for their efforts as research coordinators on this project. We also thank the Mayo Clinic Survey Research Center for their assistance with data retrieval and entry. The study was funded by a Mayo Clinic in Arizona intramural clinical research award (CR5). Salix Pharmaceuticals provided the rifaximin used in the study but was not involved in the study design, conduct, analysis of data, writing or decision where to publish.
