Ten patients with pSS were included. The Ethical Review Boards at Brest Hospital approved the study protocol (registration number NCT04206826, ClinicalTrials.gov). Diagnosis of pSS was made according to the ACR/EULAR classification criteria. Hyposialia was determined by measuring the stimulated salivary flow rate (SSF < 0.8 ml/min) and unstimulated salivary flow rate (USF < 0.3 ml/min).

The study was designed as a single-center, prospective, comparative, randomized, double-blind, cross-over controlled study. Each patient received a biofilm containing prebiotics (biofilm A) or sodium alginate (biofilm B) in a double-blind manner (operator and patient) for two treatment periods of 1 month interspersed with a 1-month washout period (Figure 1). The order of treatment (A then B or B then A) was assigned by randomization performed using the CSRandomization module of the Clinsight software. Half of the patients were randomly assigned to the group “A then B” and the other half to the group “B then A.”

Biofilm A comprised milk proteins, soy derivatives (alpha-oligosaccharide), vegetable glycerin, and water, whereas biofilm B included sodium alginate, water, glycerin, caramel, and beta carotene. Patients were required to apply the biofilm (one per day) upon waking in the morning after brushing teeth, on the gingival mucosa facing the maxillary vestibule. The patients were allowed to consume a drink until 5 min before the application but not for an hour after the application. The use of any type of mouthwash was to be avoided for the duration of the study.

The patients received a kit corresponding to each treatment period at the first visit (D0) and at the third visit (D60), including 30 biofilms (A or B) and a diary.

At each of the four visits (D0, D30, D60, and D90), data regarding the items listed in Table 1 were collected. Examination of the oral mucosa consisted of an objective clinical evaluation of the general state of the mucous membranes (e.g., redness, dryness, and degree of inflammation) with the establishment of a dry mouth clinical score (DMCS) based on the Challacombe scale (Osailan et al., 2011). In addition, at D30 and D90, the degree of redness and inflammation of the gingival mucosa near the site of application of the biofilm was assessed. The patients also completed a validated dry mouth self-report questionnaire comprising 18 items, and a self-report on specific signs and symptoms of dry mouth (such as slurred speech, chewing, swallowing, taste alterations, and burning sensations) using the VAS from 0 to 100 mm (Salom et al., 2015). As an example, 0 indicates no burning sensation and 100 indicates severe burning sensation.

Moreover, at each visit, SSF and USF rates were measured for a duration of 10 min without stimulation and 5 min with stimulation via paraffin chewing, respectively. The salivary pH was measured using a strip of pH paper.

The main evaluation criteria were tolerance to the biofilm at the end of the two treatment periods (D30 and D90), as assessed by the patients (VAS ranging from 0 to 100) and by the practitioner (degree of redness and inflammation of the gingival mucosa near the site of application of the biofilm) and self-reported adverse events related to the biofilm use.

The secondary evaluation criteria were self-assessment of oral comfort on a VAS of 0–100 mm (measurement of absolute variation at the start and end of each treatment period; absolute changes in specific signs and symptoms of dry mouth by subjective evaluation of mouth dryness; mouth burning sensation; taste alteration; chewing; swallowing, and speech difficulties, on a VAS of 0–100 mm), absolute changes in SSF and USF rates, changes in oral pH using pH paper, and alterations in the salivary microbiota assessed by molecular sequencing.

Saliva samples were collected in tubes, and total DNA was extracted using the DNeasy® Blood and Tissue Kit (Qiagen) according to the manufacturer’s instructions. The DNA was stored at −80°C until further use. PCR amplification of DNA was performed using PuReTaq™ Ready-To-Go™ PCR beads (Cytiva) according to the manufacturer’s instructions. The V1–V3 regions of the 16S rRNA gene were amplified using the primers 8F (5′-AGA-GTT-TGA-TCC-TGG-CTC-AG-3′) and 534R (5′-ATT-ACC-GCG-GCT-GCT-GG-3′) with 25 cycles of PCR at an annealing temperature of 60°C. The ZymoBIOMICS Microbial Community DNA Standard (Zymo Research) was included in the amplification repertoire as a positive control. All PCR products were resolved by electrophoresis on a 2% agarose gel in tris-acetate-EDTA buffer to confirm amplification, and subsequently sequenced with the Illumina MiSeq at the EcogenO facility (Rennes University, France).

The FASTQ files were processed with the QIIME2 software (v. 2021–8, https://qiime2.org/; Bolyen et al., 2019) and were imported in the “PairedEndFastqManifestPhred33”-format. The pipeline DADA2 was used to control the sequence quality and construct the feature table (Callahan et al., 2016).The forward and reverse sequences were truncated at 300 and 280 bases, respectively, with all other parameters set to default. Sequence count per sample ranged between 2,298 and 15,225. Prior to taxonomic assignment, reference reads were extracted using the 16S rRNA reference sequences (HOMD_16S_rRNA_RefSeq_V15.22. p9.fasta) obtained from the expanded Human Oral Microbiome Database V3 website (https://www.homd.org; eHOMD), on the basis of matches to the primer pair (8F/534R). The resulting reference reads were then trained as a Naïve Bayes classifier with the corresponding eHOMD 16S rRNA reference sequence taxonomy file for QIIME (HOMD_16S_rRNA_RefSeq_V15.22. qiime.taxonomy) that was also obtained from the eHOMD website. Core diversity analyses included alpha (Faith’s Phylogenetic Diversity; Faith, 1992) and beta diversity metrics (weighted UniFrac, and PCoA Bray–Curtis; Bray and Curtis, 1957; Lozupone et al., 2007). Each of the feature tables with taxonomic assignment at the species level was exported for further analyses. Sequence data will be made available upon request.

Quantitative variables were expressed as mean and SD, and qualitative variables were expressed as absolute value and percentage.

Although this is a primarily descriptive and non-confirmatory pilot study, the cross-over design has been classically analyzed using a mixed model including the factors treatment (prebiotic or sodium alginate biofilm, fixed effect), period (first or second, fixed effect), and order (prebiotic or sodium alginate then sodium alginate or prebiotic biofilms, fixed effect) as well as the subject factor (random effect) nested in the order factor. For exploratory purposes, the degree of significance was set at 5%.

Microbiome bioinformatics data were processed using the QIIME 2 pipeline (Bolyen et al., 2019). Non-parametric tests were used and considered significant at p < 0.05. The results were expressed as mean ± standard error of the mean. The Kruskal–Wallis pairwise test with Benjamini–Hochberg False Discovery Rate (FDR) correction was used to compare means for qualitative data related to alpha diversity indices (Benjamini and Hochberg, 1995). The PERMANOVA test was performed on beta diversity metric using the QIIME2 diversity plugin. A linear discriminant analysis of the relative abundance of taxa was performed with the linear discriminant analysis effect size (LEfSe) algorithm, where all parameters were set to default in the Galaxy web application.¹ All plots were generated through QIIME2 and Galaxy (Segata et al., 2011; Vázquez-Baeza et al., 2013).

Ten patients with pSS (one male and nine females) were included in the study, with a mean age of 58.10 ± 14.04 years. One patient reported depressive symptoms and four reported certain thyroid problems. No patient received treatment with drugs known to induce xerostomia. Two patients were ex-smokers and eight had never smoked. The mean number of alcoholic drinks consumed per day was 0.20 ± 0.42.

They were randomized into two arms of equal size: prebiotic biofilm (A) then Sodium alginate biofilm (B) group (n = 5) and sodium alginate biofilm (B) then Prebiotic biofilm (A) group (n = 5). All patients completed the study in their initial randomization group, with no missing data (Supplementary Figure 1).

Assessment of the clinical parameters at inclusion are shown in Table 2. The average USF and SSF rates were 0.3 ± 0.25 and 0.6 ± 0.39 ml/min, respectively, and the average pH was 7.3 ± 0.95. The average dry mouth clinical score, based on the Challacombe scale, was 4.7 ± 1.42, which corresponded to a moderately dry mouth. The initial clinical examination performed by the practitioner did not indicate any inflammation with normal oral mucosa in 90% of cases (n = 9) and mild inflammation in 10% of cases (n = 1). Furthermore, no erythema of the oral mucosa was observed in 80% of cases (n = 8), and mild erythema was observed in 20% of cases (n = 2). No patient presented oral candidiasis, and only one patient exhibited trauma of the oral mucosa. Regarding the self-assessment, the average VAS score for dry mouth was 58.4 ± 17.9 mm (range = 35–90 mm).

Supplementary Table 1 summarizes the responses to the self-questionnaire filled at the inclusion visit (D0). In total, 60% of the patients reported frequent dry mouth episodes, 40% reported constant nocturnal awakenings with the need to drink to reduce xerostomia, 60% complained of sometimes having thick saliva, and 80% never or sometimes experienced pain in the mouth. Regarding the chewing and swallowing aspect, 50% of the patients never had a sore throat and 50% sometimes had a sore throat when swallowing. Moreover, 50% of the patients never had difficulties in chewing and/or swallowing solid food, whereas 40% frequently or always experienced those problems. Furthermore, 70% of the patients sometimes or often needed to take sips of a liquid to swallow food, and 60% of the patients had no difficulty enjoying meals, whereas 60% had difficulties carrying on a conversation without stopping to drink.

Analysis of subjective criteria assessed on a VAS (oral dryness, taste alteration, chewing, and swallowing and speech difficulties) at inclusion (D0) and at the end of the wash-out period (D60) showed changes (Figure 2). Notably, the oral dryness VAS score of patients who received alginate sodium biofilm B in the first period (patients 1, 2, 6, 7, and 9, -dotted curves-in Figure 2A) tended to increase from D0 to D60 (Figures 2B–F). This observation demonstrates that VAS score in patients with pSS fluctuated with time.

The USF rate (Figure 2G) was not significantly different between D0 and D60 in four patients (#3, #6, #8, and #10), showed an increase in three patients (#4, #7, and #9), and showed a decrease in the three remaining patients (#1, #2, and #5). The SSF rate (Figure 2H) was not significantly different between D0 and D60 in only two patients (#3 and #6). For the other patients, upward or downward fluctuations in the SSF rate unrelated to the treatment received in the first period were noted. An increase in the salivary pH from D0 to D60 (Figure 2I) was observed for two patients (#7 and #9).

The tolerance of patients to the two biofilms during the different treatment periods is shown in Table 3. During the first treatment period, the sodium alginate (B)-prebiotic (A) arm estimated the tolerance to sodium alginate biofilm B at 80.6 ± 20.9, whereas the prebiotic (A)-sodium alginate (B) arm estimated the tolerance to prebiotic biofilm at 87.6 ± 22.6. However, during the second treatment period, the sodium alginate (B)-prebiotic (A) arm estimated the tolerance to prebiotic biofilm A at 45.8 ± 33.1, whereas the prebiotic (A)-sodium alginate (B) arm estimated the tolerance to sodium alginate biofilm B at 94.6 ± 7.4. Although the difference between the two means was not statistically significant, it appeared that the order in which treatments were received was a determinant and suggested a higher acceptability of sodium alginate biofilm (B).

Regarding the assessment of the biofilm placement site by the practitioner, only one patient showed inflammation with the prebiotic biofilm A in the first period, and it was due to a temporary trauma of the mucosa because of the rigidity of the biofilm. The tolerance to the prebiotic biofilm assessed by the practitioner was therefore 90 and 100% for prebiotic biofilm (A) and the sodium alginate biofilm (B), respectively.

Analysis of the absolute changes in the VAS scores at the beginning and end of each treatment period is shown in Table 4. The most significant result was an increase in the VAS score for mouth dryness of 10.8 ± 12.8 mm for sodium alginate biofilm B versus 0.9 ± 17.8 mm for prebiotic biofilm A. The VAS score for mouth burning sensation increased by 6.3 ± 29.2 mm with sodium alginate biofilm B and by 9.2 ± 22.2 mm for prebiotic biofilm A. The VAS scores for taste alteration and chewing difficulties remained globally comparable within the two groups. The VAS scores for swallowing and speech difficulties were improved with prebiotic biofilm A (7.9 ± 18.2 and 6.2 ± 10.6 mm, respectively) whereas no change was observed with sodium alginate biofilm B. Individual data according to the randomization group highlight the changes more precisely (Supplementary Figure 2). No change was observed in USF, and SSF rates and salivary pH regardless of the biofilm used (data not shown).

To assess the community diversity between the treatment time-points and among participants, alpha and beta diversity metrics were estimated in the QIIME2 pipeline. The alpha diversity metrics Faith’s phylogenetic diversity and Pielou’s evenness showed that there was no significant difference in richness and evenness between study visits (Figures 3A,B). The beta diversity metric weighted UniFrac indicated no significant difference in community dissimilarity between study visits were found (Figures 3C–F).

The Bray–Curtis dissimilarity emperor plots showed that study visits (Figure 4A), and study phases (Figure 4B) were clustered by participants (Figure 4C). The participants were also differentiated according to their respective treatment group (Figure 4D), and were well dispersed over the three axes displayed. Although a distinct pattern or clustering was absent (apart from clustering by participants), a slight shift from the reported baseline was observed for certain individuals (Figures 4B,C). This observation is not specific for either treatment sequence group.

Taxonomic classification was performed and presented as relative frequencies in a bar chart (Supplementary Figure 3). The 20 most abundant bacterial taxa in all samples are listed and color-coded in the figure, including the genera Streptococcus, Neisseria, and Veillonella, and species Haemophilus parainfluenzae, Veillonella atypica, Prevotella melaninogenica, and Porphyromonas pasteri. As the bar chart denotes, each patient presented a different microbial composition and slight alterations in relative frequencies were observed through the study phases. However, no significant changes were observed after each treatment, which mirrors the earlier Bray–Curtis results. Supplementary Table 2 presents a list of bacterial species and their relative frequencies across the samples.

The analyses described above suggested that the microbial community of each participant responds to the treatment phases differently, and thus, it would be difficult to discern whether there is a real effect. Despite the lack of significant differences observed in the diversity analyses, the taxonomic data at the species level was further analyzed using the LEfSe algorithm. The effect size (linear discriminant analysis; LDA) of the taxa with significantly different abundance between the groups are presented in Figure 5.

To analyze the microbiome of the study population (n = 10), the data were divided into their respective groups: the prebiotic biofilm (A) then sodium alginate biofilm (B) group (n = 5) and the sodium alginate biofilm (B) then prebiotic biofilm (A) group (n = 5). The microbiome at baseline was first compared between the two groups and no differentially abundant features was observed (data not shown). The microbial population of each treatment arm at D60 (after the washout) was then assessed to examine whether it differed compared with the microbial population at D0. No differentially abundant features were reported in sodium alginate biofilm (B) then prebiotic biofilm (A) treatment group, whereas Prevotella salivae was found to have higher abundance at D60 than D0 in prebiotic biofilm (A) then sodium alginate biofilm (B) treatment group (Figure 5A).

Moreover, the effect of the prebiotic biofilm (A) and sodium alginate biofilm (B) in the first treatment period (D0 versus D30) and second treatment period (D60 versus D90) was assessed. When prebiotic biofilm (A) was administered in the first treatment period, an increase in relative abundance was observed for the genera Prevotella (family Prevotellaceae and order Bacteroidales) and Veillonella (class Negativicutes; Figure 5B). When prebiotic biofilm (A) was administered in the second treatment period, a significant increase in relative abundance was observed for the genus Treponema (family Spirochaetaceae, order Spirochaetales and class Spirochaetia; Figure 5C). When sodium alginate biofilm (B) was administered in the first treatment period, an increase in relative abundance was observed for the genus Treponema (Figure 5D), similar to the trend observed when prebiotic biofilm (A) was used as the second treatment. However, no differentially abundant features were found when sodium alginate biofilm (B) was administered in the second treatment period. These findings suggest that the order of the treatment influenced the final outcome of the study. For example, using sodium alginate (B) as the first treatment increased the abundance of the Treponema genus. This effect persisted in the second treatment period during which prebiotic biofilm (A) was administered. Notably, Lautropia mirabilis (genus Lautropia, family Burkholderiaceae) showed a higher relative abundance at inclusion, and, Rothia aeria and family Ruminococcaceae showed a higher relative abundance at the washout (D90), prior to treatment with prebiotic biofilm (A).

When the absolute effect of each biofilm treatment was assessed, Ruminococcaceae [G-1] bacterium HMT 075 showed a higher relative abundance during non-treatment phases (D0 and D90) than during treatment with prebiotic biofilm (A; Figure 5E). However, when the absolute effect of sodium alginate biofilm (B) was assessed (Figure 5F), Fusobacterium nucleatum subsp. vincentii (family Fusobacteriaceae), Treponema lecithinolyticum (family Spirochaetaceae, order Spirochaetales, class Spirochaetia and phylum Spirochaetes), Bacteroidales [G-2] bacterium HMT 274 (from the genus Bacteroidales [G-2]), Bacteroidales [F-2] genus, Campylobacter gracilis and Leptotrichia goodfellowii showed a higher relative abundance in the treatment phases than in the non-treatment phases. Only the Mogibacterim genus showed a higher relative abundance in the non-treatment phases than during treatment with sodium alginate biofilm (B; Figure 5F). Figure 5G shows a cladogram demonstrating the effects of sodium alginate biofilm (B).

The findings showed that each biofilm had a distinct effect on the evolution of the oral microbiota in each treatment arm, in each treatment phase over the study period (Figure 6). When prebiotic biofilm (A) was administered as the first treatment, a higher relative abundance of Prevotella and Veillonella was observed. Prevotellaceae showed a lower relative abundance in the washout phase than in the first treatment period with prebiotic biofilm (A), and no difference was observed after the second treatment with sodium alginate biofilm (B). In contrast, when administered as the first treatment, sodium alginate biofilm (B) promoted an increase in the relative abundance of the Treponema genus. The increase persisted even after prebiotic biofilm (A) administration as the second treatment. Taken together, these results suggest that prebiotic biofilm (A) protected against the effects of sodium alginate biofilm (B), which appeared to promote the growth of the pathogenic genus Treponema. However, administrating prebiotic biofilm (A) as the second treatment was not sufficient to counter the effect of sodium alginate biofilm (B) as the first treatment.