Five-Year Follow-Up of Photobiomodulation in Parkinson’s Disease: A Case Series Exploring Clinical Stability and Microbiome Modulation

This was an open-label, longitudinal follow-up study of participants from a previously reported PBM proof-of-concept study of PD [24,26]. Six participants from an original proof-of-concept study, with clinically diagnosed idiopathic PD, continued PBM treatment for five years (Table 1). The remaining six participants from the original study had either been diagnosed with an alternative disease (multisystem atrophy, two participants) or had discontinued PBM treatment due to their partner passing away and relocation to an aged care facility (one participant); recovery from cancer therapy (one participant); or declining to continue PBM treatment (two participants). All continuing participants had participated in earlier PBM studies conducted under approved clinical research protocols and provided renewed informed consent. All indicated that they had continued PBM therapy with varying degrees of consistency for five years [24].

Participants self-administered at-home PBM three times weekly using a combination of transcranial, cervical, and abdominal irradiation, as previously described [26]. The abdomen and neck were irradiated using a near-infrared 904 nm Class 1 laser and one of three light emitting diode (LED) devices on the head (Supplementary Table S1). Treatment protocols were based on previously published PBM parameters shown to modulate motor and cognitive outcomes in PD and were maintained with minimal adjustment throughout the five-year period. There were a number of confounding factors that could have contributed to microbiome changes over time, such as dietary changes, aging, medications, and changes in exercise patterns. These were not directly monitored but self-reported by participants at the 5-year assessment.

Stool samples were collected from each participant at baseline and after five years of continuous PBM. Samples were immediately frozen at −80 °C until analysis. Total bacterial DNA was extracted using the Qiagen PowerSoil DNA Isolation Kit following the manufacturer's instructions, with modifications to optimise yield from low-biomass samples. DNA quality and concentration were verified by NanoDrop spectrophotometry and agarose gel electrophoresis.

All sequencing and bioinformatics analyses were performed with participant identity masked. 16S rRNA gene sequencing was chosen to maintain methodological consistency with prior analyses and as a cost-effective method for community-level profiling. The V3–V4 regions of the bacterial 16S rRNA gene were amplified using universal primers and sequenced on an Illumina (San Diego, CA, USA) MiSeq platform (2 × 300 bp paired-end reads) in two separate sequencing runs (baseline and 5 years). Raw reads were quality-filtered, denoised, and clustered into amplicon sequence variants (ASVs) using the QIIME 2 pipeline (v2024.4) with DADA2 [30]. All samples were retained after denoising. The total number of sequencing reads was 933,105, with a mean sequencing depth of 70,758. Sampling depth (33,000) was set to the shallowest sample (A2 baseline), ensuring that Shannon alpha-rarefaction had plateaued for all samples. Negative extraction and sequencing controls were included, and sequence quality metrics were assessed prior to analysis. No contaminant signatures were identified in the controls or samples; thus, decontamination tools were not applied. Taxonomy was assigned against the Greengenes2 database [31]. Alpha (α-) diversity metrics (Observed Features, Shannon index, Faith's Phylogenetic Diversity, and Pielou Evenness Index) were computed with QIIME 2 using the Kruskal–Wallis Test. Beta (β-) diversity was assessed using permutational multivariate analysis of variance (PERMANOVA), with 999 iterations for unweighted and weighted UniFrac distances visualised by principal coordinate analysis using the EMPeror plugin [32]. Changes in genus abundance between groups were tested with the Analysis of Compositions of Microbiomes (ANCOM) plugin [33] after centred log-ratio transformation. Relative abundance was determined at the phylum, family, and genus levels. At the phylum level, the various Bacillota phyla were combined to allow comparison with earlier studies. Taxa changes were exploratory, using a 2-fold difference in relative abundance as the threshold for change. Taxonomic changes were interpreted with reference to published PD microbiome signatures [9,10,11,16,34,35,36,37,38,39,40,41,42,43,44,45], with particular attention to taxa associated with SCFA production and with inflammatory potential.

All procedures complied with the Declaration of Helsinki and were approved by the Griffith University Human Research Ethics Committee (approval code 2018/16, approved 3 February 2018) with an extension until 24 April 2024. This study was registered with the Australian New Zealand Clinical Trials Registry (ANZCTR—a primary registry in the WHO International Clinical Trial Registry Platform), registration number ACTRN12618000038291p, registered on 12 January 2018. Written informed consent was obtained from all participants.

There was no significant change in the α-diversity measures of richness as determined by the Observed Features measure or Faith's Phylogenetic Diversity. However, Pielou's Evenness Index was significantly reduced (Kruskal–Wallis test, q = 0.016), as was the Shannon Index (Kruskal–Wallis test, q = 0.025) (Figure 1A). There was also a significant shift in β-diversity between baseline and 5 years for both unweighted and weighted UniFrac metrics (PERMANOVA, q = 0.003 and q = 0.036, respectively) (Figure 1B).

Changes in microbiome composition at the phylum, family, and genus taxonomic levels are shown in Supplementary Table S2. Taxonomic changes and their functional interpretations are summarised in Table 2.

Phylum-level changes (Figure 2A) in the eight most common phyla showed increases in Actinomycetota in all participants (six of six) and decreases in unclassified bacteria (five of six), Bacillota (four of six), Bacteroidota (four of six), and Pseudomonadota (four of six).

Family-level changes (Figure 2B) in the 45 most common families showed increases in Bifidobacteriaceae in five of the six participants and Ruminococcaceae in half of the participants (Figure 2C). Enterobacteriaceae, Erysipelotrichaceae, and Desulfovibrionaceae were mostly decreased in 5-year microbiomes (Figure 2C).

Genus-level changes for the 122 most common genera showed that SCFA-producing bacteria both increased and decreased over 5 years. Faecalibacterium was increased in four of the six participants in the current study (Figure 3B): in one participant (B4), increased to 13.6% of the microbiome. Other SCFA-producing bacteria, Roseburia_A_166204 and Roseburia_C, showed decreases in five of five and four of four participants, respectively, from low proportions (<1%) in most participants. Anaerostipes decreased in two of the six participants, with the remaining four showing no change. Blautia_A_141781 was overall the most common genus, representing up to 32.2% of the microbiome. Blautia decreased in two participants with lower proportions of the genus in their microbiomes (6.07% and 0.80%). Other prominent SCFA producers that have been reported as depleted in PD microbiomes (such as Butyrivibrio and Butyribacterium) were either detected at very low levels (in one participant only) or not detected, suggesting that these may have been substantially reduced or eliminated during the years of dysbiosis. Other SCFA-producing genera have been reported as being increased in PD microbiomes (Table 2), such as Bifidobacterium, which increased in four of the six participants, while Alistipes and Parabacteroides decreased in the majority of participants (Figure 3C). SCFA-producing genera that have been reported as increased in PD in some studies and decreased in other studies (Table 2) also showed either increases in the majority of participants (Butyricimonas, Prevotella, Turicibacter) or decreases (Bacteroides and the various Eubacterium genera) (Figure 3D).

Many pathobionts and pro-inflammatory genera that were detected in the 122 most common genera showed a decline in the majority of participants (Figure 3E), including Bilophila (three of four), Desulfovibrio (two of four, with two unchanged), Methanobrevibacter (two of three), and Klebsiella (two of three), while Limiplasma increased in two and decreased in two participants. Streptococcus increased in one participant and was unchanged in the other five. In each of these cases, the genera were at low levels (<1%) at baseline, apart from two participants with higher proportions of Streptococcus (8.59% and 5.32%), both of whom showed a small (less than 2-fold) decrease over 5 years. Collinsella is another pathobiont that is increased in the PD microbiome. It was the only genus to show a significant change (increase) over the 5 years, as detected by the ANCOM statistic, and was increased in five of the six participants to between 4% and 18% of the microbiome. Another conspicuous pathobiont change was the increase in Enterococcus_B in one participant (A2) from non-detected at baseline to becoming the dominant genus at 5 years (45.5% of the total microbiota).

Genera considered to be generally healthy in other contexts but that are increased in the PD microbiome include Akkermansia, Bifidobacterium, and Lactobacillus (Table 2). Akkermansia decreased in four of the six participants in the current study (Figure 3E), although one participant showed an increase in Akkermansia to 18% of the microbiome. The proportion of Lactobacillus in the samples was below the threshold for inclusion in this analysis.

There are a number of genera that have been considered to be potential markers of PD progression (Table 2). Putative bacterial markers correlated with PD severity and more rapid PD progression include Desulfovibrionaceae, Erysipelotrichaceae, Desulfovibrio, Bilophila, Limiplasma, and Eubacterium, all of which, apart from Limiplasma, showed decreases over 5 years in the majority of participants. Putative bacterial markers inversely correlated with PD progression (Bifidobacterium, Prevotella, Butyricimonas) showed increases in the majority of participants, except for Bacteroides, which was decreased in the majority of participants.

This study provides the longest reported follow-up of individuals with PD who have continued PBM therapy, extending the previous report of five-year clinical outcomes [24], to include parallel gut microbiome observations. In the earlier clinical analysis of this cohort, most participants who continued PBM therapy demonstrated sustained stability or improvement in mobility, balance, and non-motor features such as cognition and olfaction, with no serious adverse events reported. This stability or improvement in the MDS-UPDRS-III score can be contrasted with the expected decline of between 1.4 and 8.9 points annually in untreated or L-dopa-medicated PD patients [84].

Taken together with the changes in the microbiome, the findings suggest that long-term PBM was well-tolerated in this small cohort and was associated with sustained clinical stability alongside longitudinal changes in gut microbiome composition. While no causal inferences can be drawn, the co-occurrence of clinical stability and microbiome shifts provides a basis for further investigation in controlled studies.

Participants reported no major changes to their diets over the five years of this study, suggesting that large dietary changes were unlikely to account for the observed microbiome changes. Significant longitudinal changes were detected in both α- and β-diversity. The reduction in α-diversity evenness, reflected by lower Pielou's Evenness and Shannon indices, indicates increasing dominance of specific microbial taxa within the microbiome community over time. While cross-sectional studies of PD have not consistently reported differences in α-diversity compared to healthy controls (HCs), several studies have described a significant increase in the α-diversity in PD cohorts compared to HCs [35,69,85,86].

Significant changes in both unweighted and weighted UniFrac indices indicate global restructuring of the microbial community. Changes in unweighted UniFrac are consistent with shifts in the presence or absence of microbial lineages, whereas changes in weighted UniFrac suggest alterations in the relative proportions of taxa present. Significant differences in β-diversity are commonly reported in cross-sectional studies [9,56], and the present findings demonstrate that comparable compositional shifts can also occur longitudinally within individuals.

To further interpret diversity shifts, an exploratory assessment of changes in taxonomic composition was undertaken. At the phylum level, reductions were observed in the relative proportion of Pseudomonadota (formally Proteobacteria), a group frequently reported as increased in PD microbiomes and with many pathobiont and pro-inflammatory members [74]. A reduction in Methanobacteriota was also observed. This is a phylum of methane-producing archaea that are also reported as increased in PD microbiomes [9]. At the family level, the relative abundance of Bifidobacteriaceae increased. Although this family includes probiotic strains, it has been reported as increased in PD [38,51]. Decreases were observed in families containing bacteria that produce LPS and H₂S (Enterococcaceae and Desulfovibrionaceae); metabolites that have been linked to gut barrier dysfunction, inflammation, neuroinflammation, and α-synuclein aggregation [17,41,49,52,87]. Reductions were also noted in Erysipelotrichaceae, a family with strains that have been linked to gastrointestinal disruption [53] and an altered lipid metabolism [54].

Many SCFA-producing genera are reported to be decreased in the PD microbiome compared to HCs (Table 2), including the archetypal SCFA producer Faecalibacterium, which has been identified as one of the main producers of butyrate in the gut [88] and has consistently been reported as depleted in PD [56,57,58]. Enrichment of Faecalibacterium, Bifidobacterium, Turicibacter, Butyricimonas, and the Ruminococcaceae, as well as other SCFA producers, suggests increased SCFA production in the gut, although production of SCFA is strain-specific [89] and can also be affected by diet, host factors, and the gut environment [90]. The reduction of other SCFA-producing genera (Roseburia, Coprococcus, Eubacterium, Bacteroides, Gemmiger) underscores the complexity of the changes in the microbiome. The reported enrichment in the PD microbiome of taxa that are, in other contexts considered healthy (Akkermansia, Bifidobacterium, Lactobacillus) similarly suggests strain-specific and host-specific effects [38,51]. Akkermansia, for example, despite supporting mucin production and gut barrier health in some settings [15,81], has been linked to systemic inflammation and symptom progression in some PD studies [38,57] and was reduced in most participants in the present study.

In parallel with these changes in some putative beneficial taxa, there were decreases in Klebsiella, Bilophila, and Desulfovibrio, all of which are commonly enriched in PD dysbiosis [9,39,87], which suggests reduced exposure to LPS and H₂S in the microbiomes of our participants, and may indicate some stabilisation of metabolic risk pathways. A notable exception was the significant increase in Collinsella, which has been implicated in increased gut permeability and pro-inflammatory signalling [35] and is reported as associated with PD dysbiosis [36,66], PD progression [35] and Lewy Body dementia [77]. Increases in this genus highlight the need for strain-level and functional profiling, as taxa may exert context-dependent effects [35,83,91,92]. The extremely high proportion of Enterococcus_B observed in one participant at 5 years is most likely indicative of an infection or treatment with antibiotics. Enterococcus faecium, a notable species in this genus, is commensal in the gut at low levels but is also known as an opportunistic pathogen with multiple antibiotic resistances and can dominate the gut microbiome in response to broad-spectrum antibiotic exposure [93].

The microbiome findings somewhat align with cross-sectional studies that have reported depletion of SCFA producers and enrichment of pro-inflammatory taxa in PD compared to HCs. Taken together, these exploratory taxonomic changes, including increases in the relative proportion of some SCFA producers and reductions in the proportion of several pathobionts, support the hypothesis that, in parallel with clinical stability, PBM might be associated with a shift towards a more metabolically supportive and less pro-inflammatory gut environment. However, alternative explanations for the changes in the microbiome in these participants (such as diet, lifestyle, natural fluctuations, etc.) cannot be excluded.

Notably, the microbiome changes observed in this study differ from the relative stability of the PD microbiome reported in other longitudinal studies [34,94]. However, the decreases in some SCFA-producing genera, including Roseburia, and the increases in Bifidobacterium and other genera that are reported as increased in PD, together with the marked increase in Collinsella, highlight the complexity of interpreting longitudinal microbial shifts.

The mechanism by which PBM might influence the microbiome remains incompletely defined. In animal models, direct abdominal irradiation has been demonstrated in a number of studies to increase SCFA producers and reduce pro-inflammatory taxa [27,95,96,97,98,99], and direct irradiation of human faecal samples has been reported to restore cryo-damaged microbiota [100]. However, since photon penetration is limited to a few centimetres, light will not reach the interior of the gut in humans and can have no direct effect on the microbiota. Thus, interaction of PBM with the gut microbiome would be indirect. PBM has a well-known anti-inflammatory effect, directly modulating the inflammatory process by reducing pro-inflammatory cytokines (interleukin [IL]-1β, IL-6, Tumor Necrosis Factor alpha [TNF-α]) and increasing anti-inflammatory cytokines (IL-10), as well as modifying pro-inflammatory macrophages (M1) to the anti-inflammatory phenotype (M2) [101]. In addition, in animal models, PBM has been shown to modulate extracellular signal-regulated kinase (ERK), influencing the mitogen-activated protein kinase (MAPK) pathways [102,103], which in turn influences the gut-associated lymphoid tissue (GALT) and reduces local abdominal and mucosal inflammation [104,105,106]. Evidence suggests that reducing inflammation has the effect of enhancing gut barrier integrity [107]. PBM also reduces oxidative stress [104], potentially reversing damage to the gut colonic epithelial cells. Improved gut barrier integrity and reduced inflammation create an environment that favours beneficial taxa such as Faecalibacterium while suppressing LPS- and H₂S-producing bacteria. We could hypothesise that this convergence of reduced systemic inflammation, improved barrier function, and ecological shifts may go some way to explain both the observed stability in clinical outcomes and changes in microbial composition in the gut.

Other microbiome-targeting strategies in PD, such as diet, prebiotics, probiotics, synbiotics, exercise, and FMT, have produced mixed results. Mediterranean and fibre-rich diets appear protective epidemiologically but show inconsistent benefits after PD onset [108,109]. Prebiotics, probiotics, and synbiotics improve dysbiosis and reduce inflammation in animal models [110], but human trials can show selective microbial changes without a lasting clinical impact [111,112,113]. While moderate exercise is known to improve the microbiome in healthy individuals [114], evidence for this effect in PD is lacking. FMT, while highly effective in animal models, has yielded inconsistent clinical results [115,116], with a recent clinical study showing no consistent improvement in motor or non-motor symptoms with 6 months of FMT [117]. Most interventions, with the possible exception of FMT, are comparatively simple to administer. Compared with these interventions, PBM offers several advantages: it is non-invasive and well-tolerated, has a well-documented safety profile, and has shown lasting improvements in both motor and non-motor symptoms of PD in a number of studies [23,24,25,26].

In an aging non-PD population, the microbiome would be expected to deteriorate, with reduced SCFA-producing bacteria and increased pro-inflammatory bacteria [118,119]. In the PD microbiome, we would not expect an improvement in the gut microbiome over time [120]. The observed parallel clinical improvement in the symptoms of PD and the changes in the microbiome suggest the hypothesis that there may be a link between modulation of the microbiome with PBM, the gut–brain axis, and symptomatic improvement.

Future studies should include larger, well-controlled randomised trials with extended follow-up, integrating multi-omics approaches such as metagenomics, metabolomics, and immunophenotyping to link microbial function with host responses. Strain-level analysis ofandis particularly important, given their context-dependent roles. Mechanistic work on PBM's interaction with GALT and systemic inflammation and its combination with dietary fibre or probiotics may reveal additive or synergistic benefits. Akkermansia, Bifidobacterium, Collinsella