The profile of gastrointestinal dysfunction in prodromal to late-stage Parkinson’s disease

Demographic and clinical description of PD sample

A total of 404 people with PD (H&Y = 1, n = 77; H&Y = 2, n = 234; H&Y = 3, n = 57; H&Y = 4, n = 34; H&Y5 = 2), 37 people with iRBD and 72 controls participated in the study. A summary of demographic and clinical sample characteristics is presented in Table 1. When comparing across PD stages, there were no statistical differences in prevalence of iRBD history, diet, water intake and smoking. There were no differences in age, MoCA scores, opioid-based drugs and anticholinergic drugs, diet, water intake and smoking between the PD, iRBD and control groups. There were also no differences in self-reported GI diagnoses between groups except for constipation, with higher prevalences in the iRBD and PD compared to the control group. One iRBD participant was on a dopamine agonist for Restless Legs Syndrome. There was a statistically significant increase in age, disease duration, MDS-UPDRS scores, LEDD, opioid-based drugs and anticholinergic drugs across disease stage, whilst MoCA scores decreased, denoting worse disease severity across the H&Y stages.

GID symptoms across disease stage

Table 2 summarises individual and composite GIDS-PD scores across the groups. Unadjusted comparisons of GIDS-PD domain scores across the PD groups, stratified by H&Y stage, demonstrated a main effect of PD stage on GIDS-PD Constipation scores (H(3) = 36.714, p < 0.001), Bowel Irritability (H(3) = 11.402, p = 0.010) and Upper GI scores (H(3) = 50.580, p < 0.001). This indicates a general increase in self-reported GI symptoms in all domains with advancing disease stage (Fig. 1). Accordingly, we found a correlation between motor disease severity (MDS-UPDRS-III) and both GIDS-PD Constipation score (rho = 0.19, p < 0.001) and GIDS-PD Upper GI score (rho = 0.25, p < 0.0001).

Unadjusted GIDS-PD Constipation, Bowel Irritability and Upper GI domain scores increased across disease stage. GIDS-PD Gastrointestinal Dysfunction Scale for Parkinson’s disease, H&Y Hoehn and Yahr, RBD REM Sleep Behaviour disorder.

GIDS-PD Constipation, Bowel Irritability and Upper GI scores were higher in the iRBD group and PD compared to controls, but not between PD and iRBD, suggesting that the GIDS-PD is sensitive to early GI dysfunction in prodromal states. Interestingly, the iRBD group also had higher GIDS-PD Constipation than early PD (H&Y = 1), but this significance was lost after Bonferroni correction. In terms of frequency of individual symptoms, hard stools and incomplete sense of evacuation were the most reported symptoms across groups (Table 2). Interestingly, difficulty swallowing was also common across groups, with a higher prevalence in iRBD cases compared to early PD (H&Y = 1) (Table 2).

In the PD group, participants with a positive RBD history prior to PD diagnosis had significantly higher GIDS-PD Constipation scores (9.4 ± 8.0) compared to those with a negative RBD history (7.0 ± 6.4; Mann–Whitney: U = 8013.0, p = 0.016).

In the PD group, linear regression analysis was performed to further evaluate the effect of disease stage on gastrointestinal dysfunction with adjustment for relevant covariates. This showed that H&Y stage was predictive of GIDS-PD Upper GI scores but not GIDS-PD Constipation and Bowel irritability domain score (Table 3). Having a positive history of prodromal RBD was the strongest predictor of GIDS-PD Constipation scores. Exercising 3 times a week or more was also associated with decreased Constipation scores. Caffeine intake was associated with lower GIDS-PD Bowel Irritability scores and anticholinergic medication use was associated with higher Bowel Irritability scores. GIDS-PD Upper GI scores were associated with higher H&Y stage and use of anticholinergic medication. We also repeated regression analysis with MDS-UPDRS III instead of H&Y for all GIDS-PD subscores and confirmed our results, MDS-UPDRS scores were a significant predictor for GIDS-PD Upper GI scores but not Constipation or Bowel Irritability (Supplementary Table 1).

Because the prevalence of daily laxative use increased significantly across stages and including it as a covariate in the regression models would induce a bias (those with laxative use score higher in item GIDS-PD 1), we repeated the linear regression analysis for GIDS-PD constipation in participants who were not on regular laxatives (H&Y = 1, n = 67; H&Y = 2, n = 162; H&Y = 3, n = 27; H&Y = 4, n = 11) and observed similar results, with RBD prodromal history, but not H&Y and other variables, predicting GIDS-PD Constipation scores (F(6, 150) = 2.295, p = 0.038, R² = 0.09).

Longitudinal characterisation of GI scores

A subset of participants completed the GIDS-PD longitudinally. 281 participants completed it twice, with an average interval of 1.6 (±0.7) years between timepoints. Of those, 144 participants completed it a third time, on average 2.8 (±0.9) years after visit 1 and 45 completed it a fourth time, on average 3.8 (±0.8) years after visit 1. Mixed-effects analysis showed a significant increase in PD motor severity (MDS-UPDRS-III) at each timepoint (score of 29.8, 33.8, 38.8 and 45.9 respectively; F(1.213, 132.6) = 39.32, p < 0.0001) as well as in Hoehn and Yahr stage (F(1.457, 192.3) = 27.43, p < 0.0001) and LEDD (328.3 mg, 550.0.mg, 631.8 mg and 777.9 mg respectively; (F(0.7315, 91.44) = 84.17, p < 0.0001). Mixed-effects models did not show significant differences in GIDS-PD scores between timepoints over an average of 4 years of follow-up (GIDS-PD Constipation, F(2.708, 412.6) = 2.142, p = 0.1009; GIDS-PD Bowel Irritability, F(2.668, 406.4) = 1.354, p = 0.2583; GIDS - PD Upper GI, F(2.864,433.4) = 0.3570, p = 0.7748, Mixed-effects models) (Fig. 2). Participants were then stratified according to PD duration (≤2.0 years from PD diagnosis at v1, n = 162; ≥2.01 years from PD diagnosis at v1, n = 119) to investigate whether longitudinal changes in GIDS-PD scores were more prominent in early or later disease. Mixed-effects models did not reveal significant differences in longitudinal GIDS-PD scores in either group (p > 0.2852).

No significant differences in GIDS-PD scores were observed over 4 visits spanning an average of 3-year follow-up in 245 PD participants (A–C) and over 2 year follow-up (D–F) in 25 RBD cases. GIDS-PD Gastrointestinal Dysfunction Scale–Parkinson’s disease, RBD REM Sleep Behaviour Disorder. Error bars represent SD.

A paired longitudinal analyses of 40 PD participants who were drug-naïve at baseline and started on dopaminergic medication at the subsequent follow-up visit (interval between visits 1.5 y ± 0.6), showed a significant decrease in bowel frequency scores (GIDS-PD item 1, W = 133, p = 0.0416), increase in straining (GIDS-PD item 2, W = 89.00, p = 0.0502) and no differences in dysphagia or gastric emptying scores (GIDS-PD item 9, p > 0.9999; GIDS-PD item 7, p = 0.07267, respectively) between the first and second visit.

25 iRBD cases completed the GIDS-PD longitudinally, after an average time interval of 1.1 (±0.1) years and 20 completed it a third time at 2.2 (±0.4) years follow-up. At the third time-point, 3 RBD participants had converted to PD and 1 to Lewy Body dementia. Mixed effects models did not show differences in GIDS-PD scores of iRBD participants between visits 1, 2 and 3 (0.4244 > F < 1.712, 0.5417 < p < 0.2026).

Objective measurement of GID

Ninety-six participants with PD and 54 paired household controls ingested a blue food dye to measure whole gut transit time (WGTT). Ten PD cases, but no controls, took laxatives in the days following the blue dye ingestion and were excluded from the analysis. Eighty-six participants were included in the analysis (H&Y = 1, n = 27; H&Y = 2, n = 42; H&Y = 3, n = 12; H&Y = 4, n = 5). None reported discomfort following dye ingestion. There were no differences in the interval between the last normal stool and food dye ingestion between PD (0.4 ± 0.4 h) and controls (0.3 ± 0.4 h) (W = −300 p = 0.9788, Wilcoxon). WGTT was significantly higher in PD cases (62.9 ± 43.4 h) compared to household controls (36.6 ± 26.4 h) (W = −993.0, p < 0.001, Wilcoxon) (Fig. 3A). Thirty-eight PD cases (44.1%) and 8 controls (14.8%) were above normative cut-off for slow transit time (≥59 h)²⁰, with the longest transit time recorded at 240 h in the PD group, despite the participant reporting daily emptying of bowels. There were no differences in WGTT across H&Y stages (Kruskal–Wallis test, H(4) = 2.932, p = 0.4022) (Fig. 3B).

WGTT was significantly higher in PD cases compared to household controls (A). Within the PD cohort, there were no differences in WGTT across H&Y stages (B). Grid line depicts cut-off for delayed transit time (59 h). H&Y Hoehn and Yahr, PD Parkinson’s Disease, WGTT whole gut transit time. ****p ≤ 0.0001. Error bars represent SD.

Slower transit time was significantly correlated with higher GIDS-PD Constipation score (rho = 0.53, p < 0.001) and MDS-UPDRS constipation (item 1.1, rho = 0.32, p = 0.003), providing objective validation of the GIDS-PD as a tool to assess gut function in PD (Supplementary Fig. 1). WGTT correlated with age at visit (rho = 0.30, p = 0.005), LEDD (rho = 0.21, p = 0.049) and motor fluctuations (MDS-UPDRS Part IV, rho = 0.23, p = 0.036) but not with other MDS-UPDRS scores or other GIDS-PD scores. GIDS-PD constipation scores also were significantly associated with motor fluctuations (rho = 0.18, p = 0.036). There were no significant associations between WGTT and disease duration or cognitive scores. Partial correlations between WGTT and GIDS-PD Constipation scores withstood adjustment for age and LEDD (r = 0.31, p = 0.005). Between group comparison (Mann–Whitney U test) showed that PD patients with a high WGTT (≥59 h) had higher GIDS-PD Constipation scores compared to those with lower WGTT (p < 0.001), but no such differences were found for Upper GI subscores or abdominal fullness (GIDS-PD item 7), suggesting that the WGTT is more reflective of intestinal transit time rather than gastric emptying.