0
MASTERS PALLIATIVE CARE
Oral health effects of botulinum toxin
treatment for drooling: a systematic
review
Luisa Barreto Costa Corrêa
M
20201
MASTERS PALLIATIVE CARE
Oral health effects of botulinum toxin
treatment for drooling: a systematic
review
Luisa Barreto Costa Corrêa
Advisor: Professor Doctor Bernardo Sousa Pinto University of Porto, Portugal
Co-advisor: Professor Doctor Soraya Coelho Leal University of Brasilia, Brazil
2020 FA C U LD A D E D E M E D IC IN A
2 “Grandes realizações são possíveis quando se dá importância aos pequenos começos.”
3
Special thanks
I would like to thank God for His help in making this moment possible; to my husband for taking me out of my comfort zone and encouraging us to venture into a new country; to my son for making the task sweeter; to my parents for always giving me the security and love that a child desires; to Mauricio, my colleague, who accepted the challenge of helping me with my research; to my teachers Doctor Bernardo Souza Pinto and Doctor Soraya Coelho Leal for teaching and guiding me through knowledge, contributing to my academic and professional enrichment; to my employers (Court of Justice of the Federal District and Territories and State Department of Health of the Federal District in Brazil) for granting their belief in the power of enriching their employees and allowing this study break and, finally, to Portugal, a country that I will always have good memories.
4
List of abbreviations
ALS - amyotrophic lateral sclerosis BoNT - botulinum neurotoxin
BoNT-A - botulinum neurotoxin type A BoNT-B- botulinum neurotoxin type B
CFU / mL - colony-forming units per millimeters CI- Confidence Interval
CP- cerebral palsy
DAE - Dental Adverse Event DB - double blind
PD - Parkinson’s disease
RCT - randomized controlled trial S.mutans - Streptococcus Mutans
SNARE - soluble N-ethylmaleimide-sensitive factor attachment protein receptor RIMA - rimabotulinumtoxinB
5
Index
Abstract ... 7
Introduction ... 8
Salivary gland secretion and physiology ... 8
Salivary glands and secretions... 12
Salivary gland physiology ... 12
Salivary gland innervation, quality of salivary secretion and salivary flow ... 12
Salivary flow rate... 13
Dental caries and saliva ... 14
Hyposalivation and Dental Caries ... 15
Drooling and botulinum toxin ... 16
Drooling in patients under palliative care ... 16
Mechanism of action and effects of botulinum neurotoxin ... 17
Objective ... 20
Methodology ... 21
Protocol and registration ... 21
Eligibility criteria ... 21
Information sources and literature search ... 21
Study selection ... 22
Data collection process ... 23
Risk of bias in individual studies ... 23
Statistical analysis and quantitative synthesis of information ... 23
Results ... 24
Study selection ... 24
Study characteristics ... 25
Results of included primary studies ... 27
Carious lesions ... 27
6
Oral pH ... 28
S. mutans and Lactobacilli salivary counts ... 28
Risk of bias of individual studies ... 31
Discussion ... 36
Limitations... 36
Results in context... 38
Strengths and indications for future research ... 42
Conclusions ... 43
Authors’ Contributions ... 44
Declaration of Conflicting Interests ... 44
References ... 45
Appendix I - Data on salivary flow rate in the included primary studies ... 57
7
Abstract
Background: Drooling is a major morbidity of several neurological diseases. Intrasalivary botulinum neurotoxin (BoNT) injections have been used to manage this condition with most patients reporting an improvement in their symptoms. However, by decreasing salivary flow, BoNT injections may result in an increased risk of caries. Objective: To perform a systematic review in order to assess whether, in patients with drooling, intrasalivary BoNT injections are associated with increased risk for carious lesions development, and modifications on salivary composition (oral pH, buffering capacity and osmolality) and on cariogenic bacteria load.
Methods: We searched PubMed, CENTRAL, Web of Science, and Scopus for all experimental and observational studies published until May 2019. Primary study selection, quality assessment, and data extraction were performed by two independent reviewers. No studies were excluded based on their language, publication status or date of publication. Studies’ quality was assessed based on revised Cochrane Risk of Bias tool for randomized controlled trials, and on Risk of Bias in non-randomized studies tool. Meta-analysis was not performed.
Results: Searches retrieved 1025 records, of which 5 primary studies were included and analyzed. These studies included two randomized controlled trials and three quasi-experimental studies. None of the included primary studies found BoNT injections to be associated with increased risk of dental caries or with significant reductions in oral pH. One of the included primary studies reported an increase in salivary buffer capacity following BoNT. One of the included primary studies found an increase in Lactobacilli counts. As for bias, among randomized controlled trials, one was classified as of “high risk” and the other as of “some concerns”. Among quasi-experimental studies, two were at critical risk while one was judged to be at high risk of bias.
Conclusions: Currently, there is no evidence that, in patients with drooling, BoNT injections associate with increased risk of dental caries or disturbances in oral pH or salivary buffering capacity. However, the included primary studies had important limitations and differences in their methodologies.
8
Introduction
In this study, we assessed whether, by promoting hyposalivation, intrasalivary botulinum toxin (BoNT) injections associate with increased risk of carious lesions. To better understand rationale of our study, we start by reviewing salivary properties and functions, the importance of saliva’s defense mechanism and its role in the development of dental caries. Moreover, peculiarities of salivary secretions and salivary glands will be presented. Subsequently, we will review the effectiveness of BoNT in patients with drooling.
Salivary gland secretion and physiology
Saliva is a biofluid resulting from the secretion of the three pairs of major salivary glands1,2 (parotid, submandibular and sublingual), a large number of minor salivary
glands, and gingival fluid1. It is distributed mainly over the oral cavity mucosa1 and
oropharynx3–5 bathing hard and soft tissues6. Saliva is a complex secretion mixture that
contains 99% water1,7,8, inorganic6 electrolytes3 (including sodium, potassium, calcium,
magnesium6, bicarbonate, and phosphates9) and organic molecules6,10 - such as a wide
spectrum of proteins, immunoglobulins, enzymes, nucleic acids, hormones and nitrogenous products (e.g., urea and ammonia)7,8,11–13. Non-adherent oral bacteria1,7,14,
food debris1,7,14, desquamated epithelial cells1,14 and even traces of chemicals or drugs introduced in mouth can also be found in whole saliva7.
Saliva’s functions vary and include lubrication and protection, acid-base balance (buffering capacity), maintenance of tooth integrity, antibacterial activity, taste perception and digestion1,3,6,7,9,13,15,16. Lubrication is essential to mastication, swallowing, and
speech13,17. As a seromucous coating, saliva lubricates and protects oral tissues,
including the teeth7. Salivary glycoproteins - particularly mucin, the most abundant
resting saliva glycoprotein making up approximately 20-30% of whole salivary proteins18
– provide the viscoelastic character of saliva, forming a lubricating19 high film resistance13
that allows free motion of oral tissues19. Parotid proline-rich glycoprotein albumin
complex aids in the passage of food, forming a lubricating interface between teeth and facilitating chewing by incorporating into the pellicle13. In addition, the viscoelastic nature
of mucous saliva is also maintained by the secretion of various ions along with the salivary proteins6.
Saliva also acts as a barrier against irritants7, maintaining the integrity of mucous
mucosa. These irritants include, among others, proteolytic and hydrolytic enzymes7,13(elastase, collagenase and cathepsin, for example)13, carcinogens (from
9 Major salivary glycoproteins (mucins, proline-rich glycoprotein and immunoglobulins) and minor salivary glycoproteins (agglutinin, lactoferrin, cystatins and lysozyme) are involved in defending the oral cavity7,13,16. For instance, mucins’ rheological properties
provide an effective barrier against desiccation and environmental insults13; histatins -
produced in parotid and sublingual salivary glands20 - facilitate wound healing due to
their synergistic action with epidermal growth factor13,18,20 and act as inhibitors of
microbial-origin proteolytic enzymes (metalloproteases, trypsin-like enzymes, cysteine proteases) or host enzymes (collagenases)20; cysteine-containing phosphoproteins,
which are in high concentration in submandibular saliva, act against protesases protecting the mucosa7,13.
Regarding the maintenance of dental integrity, saliva is crucial in the demineralization and remineralization process7. Demineralization, a crystalline
dissolution of enamel, occurs when an oral pH of 5.0 to 5.57,21 - the critical pH range for
caries development – or lower is reached, and acids diffuse between enamel crystals7.
Subsequent dissolved minerals diffuse out of the tooth structure and into saliva7 (of note,
fluoride plays here a key protective role, as fluorapatite is less soluble than hydroxyapatite19, requiring a lower critical pH - around 4.5 – for mineral loss to occur)21.
Remineralization is the process of replacing lost minerals from the organic enamel matrix in the crystals7. For that to happen, a supersaturation of calcium and
phosphate in saliva is necessary, as they are responsible for enamel maturation and remineralization6,7,22 . Of note, above the critical pH of 5.5, saliva is supersaturated in
relation to the tooth minerals21 and as saliva provides a diffusion medium of calcium,
phosphorus, magnesium and fluoride, as well as other components into tooth enamel, it increases the hardness of tooth surface and decreases tooth permeability13. Here again,
fluoride, just like calcium and phosphate ions, keeps saliva supersaturated with respect to hydroxyapatite22; another protective function leading to the remineralization process.
Therefore, as the oral pH average ranges between six and seven7, saliva protects oral
tissues by maintaining a fluid environment with high concentrations of calcium and phosphate13.
The balance between demineralization and remineralization processes depends on salivary calcium and phosphate concentration19,23 and a supersaturated saliva with
these salts is maintained mainly by salivary proteins7. In fact, according to a systematic
review published in 2019 by Hegde et al6, salivary proteins, mainly proline-rich proteins,
mucins, histatins, cystatins and statherins attract calcium ions, promoting remineralization of the tooth surface6. For example, proline-rich proteins maintain
calcium ion supersaturation in relation to ionic phosphate in saliva19; statherin helps to
10 phosphoproteins effectively bind calcium maintaining saliva in a supersaturation state relative to calcium phosphate salts13.
The demineralization process may be also slowed by mechanisms mediated by pellicle proteins. In fact, such proteins are too large to penetrate the enamel porosities, remaining bound to the hydroxyapatite on the tooth surface and forming a protective film7,13. This allows them to allow mineral penetration into the enamel for remineralization
and to limit mineral outflow, stabilizing hydroxyapatite in dental structure7,13. Together
with salivary calcium and phosphate ions, pellicle proteins are also able to prevent the adhesion of oral microorganisms to the enamel film and slow their growth6,16. For
instance, mucin has a selective adsorption on the tooth surface and helps the initial attachment of microorganisms by promoting growth of benign commensal oral flora, and thus creating a protective barrier against acid metabolized by the cariogenic flora and, therefore, against tooth demineralization7,19.
In relation to saliva’s buffer capacity, the main mechanisms of acid-base regulation differ in the stimulated and resting states7,14. In a stimulated state, the
concentration of sodium, chlorine, bicarbonates and protein increases while magnesium and phosphorus decreases21,23. The main pH regulation system, especially during food
or drink intake, is salivary bicarbonate7,13,14,21,24, the level of which directly varies with
salivary flow rate7,13. In resting saliva, the major buffering agent is the inorganic
phosphate system7,14. The ability of human saliva to buffer acids is crucial in maintaining
oral cavity pH values above the critical pH value of hydroxyapatite dissolution, protecting teeth from demineralization25, and indirectly preventing potentially pathogenic
microorganisms colonization by avoiding optimization of environmental conditions24.
In addition to the bicarbonate and phosphate buffers7,14,21, saliva also contains
other pH-rising factors such as the protein systems7,14,22,24,25, urea and ammonia24.
Histidine-rich peptides directly act as buffers once diffused into plaque7,13. Amino acids
and peptides can be decarboxylated to form monoamines and polyamines, a process that consumes hydrogen ions13. Salivary urea, through bacterial urease14, is converted
to ammonia7,14. Altogether, these processes increase the pH of the dental plaque13,
leading to remineralization of the tooth enamel19.
Other salivary functions include digestion and taste perception. Saliva aids in bolus formation and initiates enzymatic digestion of starch.1 Salivary amylase, a
digestive enzyme, initiates the breakdown of starch7,13,26 and glycogen13,26, despite
playing an overall limited role in carbohydrate digestion7,13. Moreover, salivary lipase, a
salivary enzyme that is secreted by lingual salivary glands (von Ebner glands), initiates fat digestion7,26 playing a role in fat digestion13. For taste perception, during mastication,
11 them13,26. For instance, water and mucins help in the dissolution and transport of taste
substances to taste buds1; proline-rich proteins contribute to the sensation of
astringency1;andgustin, the zinc-specific salivary protein, is suspected to mediate taste
sensation1,7,13. Salivary hypotonicity of unstimulated saliva – with low salt, sugar,
bicarbonate and urea levels - improves taste perception as it does not compete with exogenous taste modalities7,13.
Regarding antibacterial activity, saliva contains immunoglobulins and peptides participating in innate immunity7. In addition, it interferes with bacterial adhesion and
elimination13,14,27. The antibodies present in saliva include secretory IgA14,19,21,22,27,
IgG7,14,19,22 and IgM7,14,22. IgA is the most abundant immunoglobulin in saliva, and acts by
inhibiting the binding of bacteria to host tissues7,13,19and inactivating bacterial enzymes
and toxins19. Other immunoglobulins in saliva are in small amounts and probably come
from the gingival fluid7.
Innate immune peptides include histidine and proline-rich proteins, lactoferrin7,14,21,27, lysozyme7,14,21,27, and peroxidase7,14,21,27. Parotid saliva histidine-rich
peptide, a cationic peptide, is an antifungal26 and antibacterial agent inhibiting the growth
of these microorganisms13. Such peptide destabilizes the cell membrane of bacteria
leading to cell disintegration18–20 , and its antibacterial effect covers microbes like S.
mutans20. Salivary peroxidase, converts salivary thiocyanate ion13,19 in hypocyanocyte
and hypocyanous acid, which oxidize the sulfhydryl groups of enzymes involved in glycolysis and glycid transport in bacteria13 (i.e., the products of thiocyanate and
hydrogen peroxide interaction inhibit bacterial glucose metabolism)28. In addition, the
antimicrobial effect of salivary peroxidase against S. mutans is significantly enhanced by interaction with IgA13. On the other hand, lysozyme has enzymatic activity16,19 and
interacts with low charge density chaotropic ions anions (thiocyanate, perchlorate, iodide, bromide, nitrate, chloride and fluoride) and with bicarbonate, which may cause bacterial cell lysis13 by hydrolyzing cell wall polysaccharides16, especially S. mutans13.
Finally, lactoferrin, a non-enzymatic antimicrobial protein16, is effective against bacteria
that require iron for their metabolic processes13 as it acts by chelating7,16,18.
Antibacterial activity of salivary electrolytes is also known. In addition to the interaction of some salivary proteins with calcium and phosphate, other electrolytes may act to control caries. For instance, salivary nitrate forms nitrous acid when in contact with bacteria, and this acid spontaneously decomposes to produce nitric oxide that may have a antibacterial effect19. Copper reduces enamel dissolution as it precipitates a protective
copper phosphate phase on the tooth surface and has a cariostatic effect through
12
Salivary glands and secretions
Salivary secretions are classified as serous, mucous or mixed7. Serous
secretions consist of watery saliva with little or no mucin, and are mainly produced in the parotid gland1,7,13,26,29. Mixed (serous and mucous) secretions are mostly produced in
submandibular glands1,7,29, while mucous secretions (which are richer in mucins) mainly
result from sublingual and minor glands29. Serous secretions are, thus, mainly
responsible for moistening the food, while mucous secretions coat it, facilitating ingestion13. Major salivary glands secrete most of the volume and electrolyte content of
saliva7,26, while minor salivary glands account for a small amount of secretion7,26 but for
a relative large fraction of the salivary protein mucins1. Pedersen and colleagues (2018)1,
in a review article, stated that all protein in “pure glandular saliva” came from salivary glandular cells and are not derived from blood stream. The article also stated that the majority of the 2.500 different proteins in whole saliva is likely to originate from desquamated epithelial cells and oral microflora, and only 10 per cent is thought to be of gland origin1.
Salivary gland physiology
The types of cells found in the salivary glands include secretory end pieces known as acinar cells1,7, duct system cells, and myoepithelial cells7. Acinar cells produce an
isotonic primary secretion7 with plasma-like electrolyte composition1,and determine the
type of secretion produced from the different glands (mucous, serous or mixed)7. Duct
system cells modify the saliva1 and are divided in intercalated, striated, and excretory7
cells. Intercalated duct cells are the first duct network7 and are not involved in electrolyte
variation like the remaining duct cells7. Striated cells, second in the network, act
resorbing sodium1,7 and chloride1 and in adding bicarbonate and potassium ions. Finally,
excretory duct cells continue sodium resorption and potassium secretion7. Therefore,
final secreted saliva is hypotonic26 with low sodium concentration compared to plasma1.
It is important to note that bulk protein secretion occurs in acini, but duct cells also release various proteins such as growth factors (eg, nerve and epithelial growth), immunoglobulin (IgA) and kallikrein1. Myoepithelial cells contract after stimulation to
constrict acinar1,7 and intercalated ducts1. Myoepithelial cells are controlled by the
autonomic nervous system1 and their ability of secreting or “squeezing out” saliva is a
neural process outcome7.
Salivary gland innervation, quality of salivary secretion and salivary flow
Salivary glands are mainly regulated by gastrointestinal hormones1,30 and neural
13 central nervous system circuits1,31 (particularly those related to the solitary-hypothalamic
circuit).
Major and minor salivary glands are innervated both by parasymphathetic and symphathetic nerve fibers 1,7,14,21 being the parasympathetic innervation more numerous
than the sympathetic innervation1. Depending on the nature of the stimulus, saliva
volume and composition is altered.1 In general, the parasymphathetic stimuli increase
the output of water and electrolytes whereas, when sympathetic stimuli dominate, there is an enhance of protein synthesis and secretion from acinar cells1,7,14,26.
Salivary flow rate
Healthy whole saliva flow ranges between 0.75 to 1.5 liters per day 7,26,32–35, and
stimulated saliva represents 80 to 90 percent of daily salivary production7,36. There are
several triggers that can serve as stimuli for salivary production, such as mechanical (chewing), taste (with acid being the most stimulating taste, and sweet the least stimulating), and olfact26,31. Other factors that alter the flow of secretion include pain and
other psychic factors, some medications (xerostomic or sialogogues), local or systemic diseases that affect the salivary glands7, hormonal factors, external influences and
systemic conditions37,38. For instance, previous studies have found a fluctuation in
salivary flow rate according to the circadian cycle1,7,37–39 (peaking mid-late afternoon 32,37) or even throughout the year7,21,38 (with lowest flow at summer, and peak flow at
winter 7). There is also considerable inter- and intra-individual variation32.
In the unstimulated state, the parotid glands are responsible for 20% of salivary production, the submandibular glands account for 65% of that production, the sublingual for 7-8%, and smaller glands for less than 10%7. In the stimulated state, the parotid
glands contribute with more than 50% of secreted saliva7,40.
Unstimulated salivary flow (“resting saliva”) lies around 0.25-0.35 mL/min1,14,21
while stimulated salivary flow ranges from 1 to 3 mL/minute 1,14,36. Unstimulated flow rates
below 0.1 mL/ min 1,7,14 and stimulated flow rates below 0.50-0.70 mL/ min are considered
hyposalivation1,14,22.
Salivary flow is not equally distributed throughout the oral cavity.7,21 The mandible
is a high volume site, while the anterior maxillary sites and interproximal regions are low volume flow sites7. This is important because some studies have already pointed out that
the regional clearance rate of acid produced from bacteria is directly influenced by regional variations in oral cavity flow7,40 and, where salivary flow is lower, acid
by-products may remain in longer contact with oral structures7.
Oral pH7,13,37,41 and salivary osmolality also vary with salivary flow. Regarding oral
14 to 7.8 (peak flow conditions)7. Regarding salivary osmolality (salivary concentration
expressed in milliosmoles of solute particles per kilogram of water42), it reflects the
hydration status42–44, with increased salivary osmolality being associated with
dehydration42,44and significantly correlating with low salivary flow rate42,43.
Dental caries and saliva
Saliva is important to prevent imbalances of the oral environment and their consequences, including changes to the existing microbiome45.
Dental caries is a polymicrobial46,47 and multifactorial disease6,27 considered one
of the most common health problems5,27,48–50. Factors contributing for caries formation
include microbiological changes inside the biofilm complex23, frequent dietary
carbohydrate intake45, salivary flow and composition, low exposure to fluoride, and
insufficient teeth cleaning23. A direct correlation of hyposalivation and dental caries is
observed27,43,45.
The dental plaque is a dynamic microbial ecosystem23,46,51 , whose predominant
bacteria include Veillonella23,27, Actinomyces23,27, Streptococcus23,27, Neisseria and other
aerobic organisms23. Bacteria exist on teeth in micro-colonies structures made up of
proteins, carbohydrates and nucleic acids exported by the cells23. These components
protect bacteria from host defense, dehydration, and increases resistance to antimicrobial agents23.
Dental caries results from an ecological dysbiosis50. The biofilm on clinically
sound enamel surfaces mainly contain bacteria other than S. mutans (“non-mutans bacteria”), in which acidification is mild and infrequent, maintaining dynamic stability51.
Under those normal and regular conditions, symbiotic microbial communities in human teeth are observed, mainly composed of Gram-positive saprophytic bacteria27 such as
S. mitis, S. oralis, S. sanguinis, S. salivarius and S. gordonii. When sugar is frequently supplied or when salivary secretion is too low to neutralize the produced acids, there is a decrease in oral pH.5,51 This acidification enhances the acidogenicity and acidurance
of the non-mutans bacteria, resulting in the establishment of a more acidic environment.46,51 This prompts a shift of the demineralization/remineralization equilibrium
to an acidogenic stage, resulting in net mineral lossesat tooth surface51. If this severe
and prolonged acidic environment continues, an aciduric stage consolidates with aciduric bacteria becoming dominant through induced selection by temporary acid-impairment and acid-inhibition of growth.51 At this phase, S. mutans, Lactobacilli51–53,
aciduric strains of non-mutans streptococci, Actinomyces, bifidobacteria, and yeasts may become prevalent51, being S. mutans and Lactobacilli more competitive under severely
15 acidic conditions51,53.This shift in microflora composition leads to the demineralization of
tooth surfaces and initiation and progression of dental caries51.
S. mutans metabolizes sucrose at a much lower pH (4.0) than most other oral microorganisms21 and converts it to lactic acid. This further promotes a drop in salivary
pH which favors an increase of Lactobacilli more than of S. mutans populations54. Some
studies referred that bacterial counts of S. mutans and Lactobacilli can be used as salivary biomarkers of caries lesions19,22, with S. mutans being associated with high
susceptibility to dental caries51,55 and with the onset of the disease22,51,56, whereas
Lactobacilli counts signal caries progression 9,22,51,57,58 as the latter were opportunists
favored by the environmental shift51.
Of note, it is important to highlight that other synergies between different bacterial species also appear to be important in caries pathogenesis.50 For instance, mixed
cultures of S. mutans and Veillonella alcalescens were also found to produce higher acid levels than biofilms containing only one of these species.50 Yet, it seems that each lesion
appears to harbor a different combination of bacteria with the estimated bacterial diversity being lower in enamel caries lesions and higher in "open" dentine lesions, supporting the concept that consortia formed by multiple microorganisms act collectively to initiate and expand the cavity50.
Hyposalivation and Dental Caries
Decreased salivary flow ratehas been identified to be related to increased risk of tooth decay27,43,45. In fact, a decrease in salivary secretion rate leads to a decrease in
salivary bicarbonate (HCO3), pH, and buffering capacity, as well as to lower clearance of
microorganisms and less removal of food debris and sugars51 from the diet in the oral
environment, favoring caries activity8,9,25. If salivary flow remains low over time, the
protective effects of saliva are reduced and the individual may be at higher risk for caries2,19,26,27. For instance, low salivary flow rates and increased cariogenic bacteria
counts have been observed in patients that had made radiation treatment51, who took
salivation-reducing medications9 or who have primary Sjögren's syndrome25.
Of note, there is often confusion between the concepts of hyposalivation and xerostomia. While they are frequently used as synonymous, xerostomia is defined as a persistent sensation of dry mouth1,14,59, and hyposalivation corresponds to a low saliva
16
Drooling and botulinum toxin
Hypersalivation, sialorrhea and drooling are terms that are often used wrongly2.
Hypersalivation refers to increased production of saliva2, while drooling is defined as the
inability to control oral secretions, resulting in the involuntary loss of saliva and its components from the mouth61, as well as in excessive saliva accumulation in the
oropharynx62,63. Drooling could be referred as intractable sialorrhea as well64. Sialorrhea
should refer to (i) primary sialorrhea, when there is excess production of saliva, and (ii) secondary sialorrhea, consisting of the accumulation of secretions due to impaired neuromuscular control1. There are also the concepts of "anterior drooling"65 or anterior
sialorrhea2, used to characterize unintentional loss of saliva dripping over the lip65, and
of “posterior drooling”65 or posterior sialorrhea2, used when saliva falls posteriorly at the
top of the tongue and pools in the hypopharynx, posing a high risk of aspiration65.
Drooling in patients under palliative care
Drooling is a major morbidity in several neurological diseases.2,66–69 Therefore,
palliative care patients, especially those with neurodegenerative conditions or neurological disorders - such as cerebral palsy (CP), amyotrophic lateral sclerosis (ALS), Parkinson’s disease (PD), multiple system atrophy and corticobasal degeneration among others - can benefit from treatment to reduce drooling63,70,71 . In fact, the
prevalence of sialorrhea is high in patients with stroke (affecting up to 80% of such patients)2,61,64,65,67, acquired brain injury2,64, and neurodegeneratives diseases such as
motor neuron disease2,61,64,65,72(up to 50%)65,67, PD2,61,64,65,67 (up to 74%)64,65,67, CP2,64,65
(up to 40%)65,73, progressive supranuclear palsy69,74, multiple sustaining atrophy69,74 and
corticobasal degeneration69. Sialorrhea generally worsens as diseases progress69.
Patients with cognitive impairment, dementia, facial palsy, laryngectomy, and post-mandibulectomy can also suffer of sialorrhea61.
The causes for sialorrhea can differ depending on the underlying disease. For instance, in motor neuron disease, degeneration of bulbar neurons causes weakness in the orofacial and lingual muscles, which can result in difficulties in eliminating oral secretions, leading to perioral ulcerations and risk of aspiration pneumonia2. In PD, the
drooling mechanism is multifactorial and includes reduced salivary swallowing and facial hypomimia2,75. For CP, excessive drooling is associated with severe motor limitations
and intellectual impairment2.
Drooling usually results in maceration and infections of the peri-oral skin2,61,65,67,
dysfunctional eating61,64, disturbed speech2,61,64,67,76, wetting and damaging technical
17 others. Neurologically impaired individuals with severe presentations may even require intravenous fluid replacement77. Drooling also associates with stigmatization or social
isolation2,17,64,67, as well as with hygienic problems for caregivers because of constant
soiling of clothes2, computers, furniture, and toys61,78. Therefore, managing sialorrhea is
crucial to diminish the overall burden on both patients and their caregivers2.
Several therapeutic options have been proposed to manage drooling, from conservative non-invasive approaches (such as oral-motor therapy, speech therapy and pharmacological treatments that attempt to decrease salivation) to invasive methods (including surgery with resection of the submandibular glands, resection of the chorda tympani nerve, and tympanic plexus neurectomy)2,61,71,79–83. However, there is a lack of
evidence supporting the use of conservative treatments, which can additionally be poorly tolerated66 (some anticholinergic drugs are frequently associated with important side
effects61,67,68,83, such as cognitive impairment, drowsiness and urinary retention82) or can
be unfeasible and unsuitable for patients with progressive neurological conditions2.On
the other hand, invasive procedures are often impossible to be performed in palliative care patients, notably when they are not able to tolerate such procedures. In fact, surgery in patients with very poor health can be considered a disproportionate measure84.
The need for effective and safe approaches to manage drooling prompted the development of new therapeutic options. One of such methods consists of injecting BoNT into the salivary glands.
Mechanism of action and effects of botulinum neurotoxin
BoNT comprises proteases (seven serotypes from A–G)85,86, and is produced by
Clostridium botulinum bacteria85,86. This toxin is used for manage various medical
conditions, including bruxism87, temporomandibular dysfunction88, peripheral
neuropathic pain89, cervical dystonia86, pelvic floor myofascial pain90 and urinary
incontinence91. In addition, it may have aesthetic uses92. BoNT has been shown to be
effective in the management of sialorrhea69,93 with several clinical trials94–100 in
amyotrophic lateral sclerosis, CP and PD patients showing that BoNT injections are tolerable and effective, with most patients reporting a transient improvement in their symptoms93. In fact, those trials have consistently observed BoNT use to be associated
with improvement in quality of life questionnaires, Clinical Global Impression scale, Drooling Frequency and Severity Scale or salivary flow rate85.
BoNT is set to reduce saliva production by a prolonged temporary inhibition of autonomic innervation to salivary glands, namely through blocking of parasympathetic2
and postganglionic sympathetic acetylcholine release63,66,71, albeit without causing
18 sympathetic nerve endings in the glands101,102. BoNT cleaves67 and inactivates68 a
soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE)68 – a
protein complex - inhibiting vesicular docking, and reducing the release of acetylcholine at the parasympathetic nerve terminals within the salivary gland.67 The different types of
BoNTs cleave different proteins inside SNARE2,68, being BoNT type A (BoNT-A) and
BoNT type B (BoNT-B) popular first-line treatment options65,67,103 for chronic sialorrhea.
BoNT-A cleaves SNAP25 (synaptosomal nerve-associated protein 25)86, whereas
BoNT-B cleaves VAMP/synaptobrevin protein4,68,76,95.
Besides the differences regarding the molecular interactions of BoNT-A and BoNT-B, the preparation of the commercial vial varies and the effective potency of the units of the four commercially BoNT preparations do not have a comparable effect in patients104. A general conversion estimation is to consider is 1 Unit (U) of
OnabotulinumtoxinA as equivalent to 1 U of IncobotulinumtoxinA, to 3–5 U of AbobotulinumtoxinA and to 50–100 U of RimabotulinumtoxinB (Rima)104. BoNT-A dose
generally varies between 100 U64,85,105 and 250 U85,94,106 and BoNT-B dose between 2500
U67,85,106 and 3500U67. Nevertheless, studies compared 250 U of AbobotulinumtoxinA
with 2500 U of rimabotulinumtoxinB and concluded that both have similar efficacy, safety and duration in managing sialorrhea106,107, although BoNT-B has a lower latency106.
Regarding the application method, the use of ultrasound as a method of guidance instead of anatomical landmarks is still debatable in the literature, but it is generally recommended, since it is safe and relatively easy to perform85. And, because BoNT
effect lasts until the presynaptic terminals regenerate - usually up to three months85,102,
repeated injections are required85,108. In fact, studies assessing the effectiveness of
BoNT have presented with a variable follow-up period67,83,108, the maximum being 64
weeks (with repeated injections of BoNT each 3 or 4 months)83.
However, by prompting a decrease in salivary flow, botulinum toxins may have important adverse effects and may be associated with changes in the salivary composition101,102 and dental caries risk2,67. There are reports of tooth loss2,64 and dental
caries2,67 after injecting BoNT into the salivary glands to treat sialorrhea2. As previously
discussed, pathologically low levels of salivary flow are associated with increased risk of dental caries2,22.
Evidence on the risk of caries following BoNT injections has not been systematically reviewed. Therefore, it is important to assess whether decreases in salivary flow subsequent to BoNT use lead to substantial changes in saliva, increasing the risk of caries. This evaluation can avoid further complications in oral health as an
19 accurate caries risk assessment identifies patients at high caries risk and encourages preventive therapies and greater treatment effectiveness5.
20
Objective
The objective of this study is to perform a systematic review of experimental and observational studies in order to assess whether, in patients with drooling, BoNT injections into the salivary glands associate with increased risk of dental caries, modifications on salivary composition (including salivary pH value, buffering capacity of saliva and osmolality) and modifications on counts of cariogenic bacteria, including salivary counts of S.mutans and Lactobacilli. In addition, this study aimed to assess the methodological quality of existing evidence on the safety of intrasalivary BoNT for treatment of drooling, discuss the main limitations of the current evidence, as well as to produce methodological recommendations for future studies on this field.
21
Methodology
Protocol and registration
This systematic review was conducted in accordance with the "Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement109 and
has been registered on PROSPERO database (Registration CRD42019137023).
Eligibility criteria
We included experimental and observational studies in which intra-salivary botulinum toxin injections were used to treat drooling irrespectively of patients’ underlying disease. We included studies that reported on adverse events of BoNT on oral health, particularly regarding (i) the risk of caries; (ii) modifications of salivary composition (salivary pH value, buffering capacity of saliva, and hydration level - osmolality); and (iii) modification of salivary counts of cariogenic bacteria, especially Streptococcus mutans and Lactobacilli. No studies were excluded based on their language, publication status or date of publication.
Information sources and literature search
We searched PubMed, CENTRAL, Web of Science and Scopus on May 2019 to identify relevant primary studies. Manual searching was also performed to collect data reported in books and conference abstracts, as well as by searching references of relevant published studies identified in electronic database search.
22 Table 1. Search queries
Databases Search query
PubMed and CENTRAL #1 sialorrh* OR sialorrhea OR hypersaliv* OR drool* OR salivar* OR salivat* OR saliva OR watering OR oversalivat* OR ptyal* OR watery mouth OR hypersecret* OR sialosis OR polysialic OR sialis
#2 MeSH descriptor: [Sialorrhea] explode all trees
#3 botulin* OR botulinum neurotoxin OR clostridium botulinum OR botulinum toxin OR botox OR dysport OR “Myobloc” OR rimabotulinumtoxin* OR onabotulinumtoxin*
#4 MeSH descriptor: [Botulinum Toxins, Type A] explode all trees #5 streptoco* OR lactobacil* OR cariogen* OR bacter* OR microflora OR microorganism OR cavit* OR carie* OR oral OR dental OR enamel OR dentin OR demineraliz* OR tooth OR teeth OR mouth*
#6 MeSH descriptor: [Dental Caries] explode all trees #7 MeSH descriptor: [Oral Health] explode all trees #8 MeSH descriptor: [Bacterial Load] explode all trees #9 (#1 OR #2) AND (#3 OR #4) AND (#5 OR #6 OR #7 OR #8)
Web of Science ALL=(sialorrh* OR sialorrhea OR hypersaliv* OR drool* OR salivar* OR salivat* OR saliva OR watering OR oversalivat* OR ptyal* OR watery mouth OR hypersecret* OR sialosis OR polysialic OR sialis) AND ALL=(botulin* OR botulinum neurotoxin OR clostridium botulinum OR botulinum toxin OR botox OR dysport OR "Myobloc" OR rimabotulinumtoxin* OR onabotulinumtoxin*) AND ALL= (streptoco* OR lactobacil* OR cariogen* OR bacter* OR microflora OR microorganism OR cavit* OR carie* OR oral OR dental OR enamel OR dentin OR demineraliz* OR tooth OR teeth OR mouth*)
Scopus (sialorrh* OR sialorrhea OR hypersaliv* OR drool* OR salivar* OR salivat* OR saliva OR watering OR oversalivat* OR ptyal* OR watery mouth OR hypersecret* OR sialosis OR polysialic OR sialis) AND (botulin* OR botulinum neurotoxin OR clostridium botulinum OR botulinum toxin OR botox OR dysport OR “Myobloc” OR rimabotulinumtoxin* OR onabotulinumtoxin*) AND (streptoco* OR lactobacil* OR cariogen* OR bacter* OR microflora OR microorganism OR cavit* OR carie* OR oral OR dental OR enamel OR dentin OR demineraliz* OR tooth OR teeth OR mouth*)
Study selection
After removing duplicates, two reviewers (LC and MB) were independently involved in selecting the studies firstly by title and abstract reading, and then by full text reading. All steps were performed under the guidance of the third author (BSP).
23
Data collection process
Two researchers (LC and MB) independently collected data from primary studies, using specifically designed forms. A pilot version was firstly built, and subsequently adapted after assessment of the first three primary studies. Extracted data from primary studies included the number of patients and their sex and age distribution, underlying/previous disease, oral hygiene, values of salivary flow rate, botulinum toxin type, administration dosage of botulinum toxin, number of botulinum toxin sessions, administration dosage of the drug, site of application of botulinum toxin, associated surgery or other therapy, follow-up time, salivary pH value, buffering capacity of saliva, osmolality and salivary counts of the cariogenic bacteria, especially S. mutans and Lactobacilli. In the absence of important and relevant data, the authors of the included studies were contacted for providing additional information.
Risk of bias in individual studies
Two researchers (LC and MB) independently assessed the risk of bias of included primary studies, with the risk of bias of each outcome of interest being separately evaluated. For assessing randomized trials, we used the "Revised Cochrane risk-of-bias tool for randomized trials"110 (RoB 2.0 tool). RoB 2.0 evaluates the risks across five
domains, namely bias (1) arising from the randomization process, (2) due to deviations from intended interventions, (3) due to missing outcome data,(4) in the measurement of the outcomes, and (5) in the selection of the reported results110.
For non-randomized studies, the risk of bias was assessed using the "Risk of Bias in Non-randomized Studies - Interventions (ROBINS-I)"111 questionnaire. This tool
evaluates the risk of bias across 7 domains, namely bias (1) due to confounding, (2) in the selection of participants, (3) in the classification of interventions, (4) due to deviations from intended interventions, (5) due to missing data, (6) in the measurement of outcomes, and (7) in selection of the reported results111.
Statistical analysis and quantitative synthesis of information
Results were presented using descriptive statistics. We performed a Chi square test to assess the statistical significance of the incidence of carious lesions in the Mysticol study67 (by using the page http://quantpsy.org/chisq/chisq.htm). A meta-analysis was not
performed due to the small number of included studies, and to the important methodological and clinical differences between them.
24
Results
Study selection
Figure 1 provides an overview of the PRISMA flow diagram of the study selection process. We retrieved a total of 1025 records through database searching and nine through manual search. After removal of duplicates (n=254), 780 studies were screened by title and abstract reading. Eighty articles were fully read, out of which 4 primary studies were selected and included in this systematic review 41,55,94,101 after database search.
Another study, “The Mysticol study”67, a multicenter randomized clinical trial was included
in this systematic review after manual search – its description was published in conference abstracts112–115, on a poster116, on an article suplement117 and at
clinicaltrials.gov under the identifier NCT01994109118 (data updated until September 10,
2019).
25
Study characteristics
Two studies included in this review were randomized controlled trials (RCT)41,67
and three were quasi-experimental studies55,94,101.
One of the RCTs - “The Mysticol study”67 (Stuart Isaacson et al) - was a phase 3
multicenter trial conducted from November 2013 to January 2017 at 33 sites in the United States, Ukraine, and Russia. It was a 13-week follow-up trial that evaluated the efficacy and safety of bilateral rimabotulinumtoxinB (RIMA) injections at 2500U and 3500U doses on the parotid and submandibular glands in the management of sialorrhea. This double-blind study included adults aged 18 to 85 years old with various types of underlying diseases and compared RIMA 2500 U, RIMA 3500 U or placebo. Ultrasound-guidance for injections occurred whenever possible. According to existing reports67,115, patients
underwent scheduled dental exam by an independent dentist at week 4 and week 13. The protocol trial reports also that an oral examination would be performed by the investigator at each study visit, where a dental examination was not planned by protocol. “The Mysticol study”67 double-blind phase was followed by an open-label extension to
determine the efficacy, safety and tolerability of repeated injections. The open-label phase results of this study and those of another open-label study from the same research group will be published in the near future according to the published report on 2020 Mysticol study article67.
The other RCT41 is a Taiwanese double-blind pilot study published in 2011 by Wu
et al. It does not provide information regarding when and where sample and data collection were carried out and its research protocol record was not found. Despite the manual search in the main databases of clinical trial records and attempts to contact the authors, no information has been provided. This trial evaluated, throughout a 12-weeks follow-up, the therapeutic effect of weight-dependent low-dose ultrasonography-controlled BoNT-A bilateral injections on parotid and submandibular glands in children with cerebral palsy (3 to 16 years old), assessing salivary composition and cariogenic bacterial counts.
For the quasi-experimental studies, the most recent one was conducted in Estonia (2018)94 and performed in adults aged 58 to 88 years old. It aimed to determine
changes in the oral microflora and saliva in patients with Parkinson disease treated for sialorrhea. Bilateral sonography-controlled injections of 250 U of BoNT-A were applied in the parotid and submandibular glands. Patients who received BoNT-A injections were compared with two control groups who did not receive any intervention – (i) a group of Parkinson diseases patients without sialorrhea, and (ii) healthy age-matched participants. The follow-up time for salivary assessments was of 4 weeks.
26 Another quasi-experimental study conducted by the same research team (Tiigimäe‐Saar (2017))55 aimed to assess the therapeutic effect of ultrasound-controlled
low dose BoNT-A injections into the parotid and submandibular glands. This study included children and adults with sialorrhea subsequent to underlying diseases and did not recruit participants for control groups to make a comparison. It was a before-and-after study, with participants at the start of the study being compared with themselves at the end of the study.
Finally, the third included quasi-experimental study was conducted in Denmark by Moller et al (2015)101 and assessed children (9-16 years old) with cerebral palsy (CP)
and disabling drool. This study applied injections of BoNT-A into the submandibular and parotid glands, aiming to identify the lowest effective dosage. For this, the authors performed six series of increasing doses, evaluating the response of each one to each dose for up to 20 weeks in each intervention. The interval between each intervention series was six months in an attempt to wash out the effect of the previous intervention. It was another before-and-after study, meaning participants at the start of the study were compared to themselves at the end of the study.
The characteristics, demographic data and summary of studies included are presented in Table 2; interventions and outcomes of included studies are summarized in
Table 3. Table 4 presents a before-and-after comparison between intrasalivary injections between the groups that received botulinum toxin injections. The data regarding the salivary flow values are presented in the table in Appendix I.
27 Table 2. Characteristics, demographics and summary of included studies Author Country Study
Design Number of participants Underlying Disease (%) Age [mean years old (range)] Gender (male/female) Stuart Isaacson et al. (2020)
Multicenter RCT 187 (184 for mITT
population) PD (65%), stroke (7%), ALS (6%), medication-induced sialorrhea (3%), adult cerebral
palsy (2%) and other disorders (16%) Adults [64(18-85)] 147/40 (144/40 for mITT population) Tiigimäe‐ Saar et al. (2018) Estonia Quasi-experimental study 38 PD (31% with and 34% without sialorrhea) Adults [71(58–88)] 22/16 Tiigimäe‐ Saar et al. (2017) Estonia Quasi-experimental study 20 PD (60%), ALS (15%); birth hypoxia (10%); atypical headache (10%) and stroke (5%) Adults and children [63(3-79)] 12/8 Moller et al. (2015) Denmark Quasi-experimental study 14 CP Children [9(5-16)] 8/6 Wu et al. (2011) Taiwan RCT 20 CP Children [8(3-16)] 9/11
ALS: Amyotrophic lateral sclerosis; CP: Cerebral Palsy; DB: double-blind; ꭞITT: Intention to treat population; mITT: modified Intention to treat; OL: Open Label; PD: Parkinson’s disease and RCT: Randomized controlled trial.
Results of included primary studies
Out of the five included primary studies, salivary buffering capacity was assessed by two55,94; oral pH and S. mutans and Lactobacilli salivary counts were assessed by
three41,55,94 and carious lesions were assessed by two67,101.
Carious lesions
In the double blind (DB) phase of the multicenter RCT by Stuart Isaacson67,
carious lesions were observed in 8% of the patients of the 2500 U (5/63), 5% of the patients of the 3500 U group (3/64), and 3% of the patients of the placebo group (2/60). From these data, we were able to calculate a relative risk of developing new carious lesions (compared to placebo) of 2.4 (95% confidence interval [CI]=0.48-11.80) for the 2500 U group and 1.4 (95%CI=0.24-8.12) for the 3500 U group. The results of the
Chi-28 square test performed by us did not show statistically significant differences in group comparison (p=0.504).
Carious lesions were also assessed in the Danish open-label study of Moller et al101, which reported no cases of tooth decay.
Salivary buffering capacity
The Estonian studies of Tiigimäe‐Saar et al found different results with respect to change in buffering capacity. For the analysis carried out in 201755 no statistically
significant change was observed, while it was detected at 2018 analysis (with increased buffering capacity one month after BoNT injections)94.
Oral pH
Wu et al41 did not observe significant differences in participants’ oral pH before
and after BoNT-A injections between the active and control groups. There were also no statistically significant differences regarding salivary pH in the two other quasi-experimental studies55,94 also that evaluated this variable.
S. mutans and Lactobacilli salivary counts
The RCT by Wu et al41 did not provide any data - either in the form of primary
data or in a measure of effect - related to cariogenic bacterial count, only presenting the hypothesis tests results (in the form of p-value) for the comparison between the baseline and the post-intervention period. The authors reported no statistically significant changes in S. mutans and Lactobacilli CFU counts. The studies of Tiigimae-Saar did not find statistically significant differences in S. mutans CFU counts either55,94. However, the
2018 study found that Lactobacilli CFU counts were significantly increased one month after BoNT injections94.
29 Table 3. Interventions and outcomes of included studies are summarized
Author Intervention Comparision submandibular/parotid Dose units per
gland respectively Outcomes Follow-up time Key findings
Stuart Isaacson
et al. (2020)
DB phase: Bont-B (Myobloc®) 2500 U and Bont-B (Myobloc®) 3500 U; OL phase: Bont-B (Myobloc®) 3500 U at first
OL cycle (dose adjustments allowed). Bilateral injections into parotid and
submandibular glands DB phase: Volume-matched placebo. OL phase: None. 2500 U group: 250/1000 U; 3500 U group: 250/1500 U Dental Caries DB phase: 13 Weeks; OL phase: every 13 weeks (maximum of 4 treatment sessions post-injection periods
Dental caries in DB phase respectively: 2500U group: 8%; 3500U group: 5%; placebo: 3%
Dental caries at OL phase Group -cycle 2: 29% (p=0.504)
Tiigimäe-Saar et al. (2018)
BoNT-A (Dysport ®) bilateral injections into the parotid and submandibular glands
Participants’ pre and post-injections salivary composition and cariogenic
bacterial counts in intervention group and between intervention and
control groups. 250 U (total dose) Buffering capacity; oral pH; SM and LB salivary levels 4 weeks No statistically significant change in oral pH (p = 0.687)**
and SM CFU count groups (p=0.206), but buffering capacity (p = 0.037) and LB CFU
counts (p=0.047) were increased Tiigimäe-Saar et al. (2017)
BoNT-A (Dysport®) bilateral injections into parotid and submandibular glands
Participants’ pre and post-injections salivary composition and cariogenic bacterial counts.No control
group. Weight-dependent dose α Buffering capacity; oral pH; SM and LB salivary levels 4 weeksᵜ No statistically significant change in oral pH (p= 0.494), buffering capacity (p= 0.082), SM (p= 0.619) CFU count groups and LB (p= 0.054) Moller et al. (2015)
BoNT-A (Botox®) injections in six successive increasing dose series. Bilateral injections
into parotid and submandibular glands
Participants’ pre and post-injections UWS flow and composition parameters Series A:10/0 U; Series B:15/0 U; Series C:20/0 U; Series D:20/20 U; Series E:30/20 U; Series F: 30/30 U
Dental Caries 20 weeks No reports of dental Caries
Wu et al.
(2011) BoNT-A (Botox®) bilateral injections into parotid and submandibular glands Normal saline placebo injections Weight-dependent doseᵠ
Oral pH; SM and LB
salivary levels 12 weeks
No statistically significant change in oral pH (p= 0.398)
and LB and SM CFU count groups (no data available)
Bilateral sublingual gland excision (BSLE); Bilateral submandibular duct relocation (BSDR); Botulinum neurotoxin (BoNT); Clinical Global Impression Change (CGI-C) ; double-blind (DB); Lactobacilli (LB); Open Label (OL) ;Streptococcos mutans (SM); Unit(U); Unstimulated Salivary Flow Rate (USFR); Unstimulated whole saliva (UWS); αThe recommended dose for Botox® is 1.4 U/kg to each parotid and 0.6 U/kg in each submandibular gland.
The recommended Botox® dose was tripled to obtain a relevant dose of Dysport®; ᵜfollow-up 1, 2, and 3 months for others outcomes but only 1 month for salivary tests; ᵠ Total dose was body weight titrated, using 30 U for subject weights <15 kg, 40 U for subject weights from 15 to 25 kg, and 50 U for subject weights >25 kg. Maximum dose for each submandibular was 10 U, and no participant received more than 50 U in total. ** Wilcoxon test. Results different from those published in the article.
30 Table 4. Comparison between before and after intrasalivary botulinum toxin injections. Author Number of subjects in intervention group S. mutans Counts median(range) Lactobacilli Counts median(range) Oral pH mean(±SD) Buffering Capacity mean(±SD) Before BoNT 1 month after Before BoNT 1 month after Before BoNT 1 month after Before
BoNT 1 month after
Tiigimäe‐ Saar et al. (2018)¥ 12 2(0–3) 2(2–3) 1(0–3) 2(0–3) 6,86* (±0.56) 6,93* (±0.78) 6,25 (±3,14) 8,83 (±2,67) Tiigimäe‐ Saar et al. (2017)¥ 20 2(0–3) 2(0–3) 1(0–3) 2 (0–3) (±0.7)6.7 (±0.7)6.8 (±3.2)6.5 (±3.1)8.2 Wu et al. (2011) 10 Measured but no data providedⱷ Measured but no data providedⱷ 7.0 (±0.2) 7.1 (±0.1) Not measured
¥Microbial tests were evaluated by counting colony-forming units per millimeters (CFU/ml): class 0 = < 103; class 1 = 103–104; class 2=104– 105; class 3= > 105. ⱷNo data available. There was only information "no significant difference was found in changes in S mutans and Lactobacilli colony count levels 1 month and 3 months after injection between groups. Similarly, no significant difference (p=0.398) between groups in oral pH (being pH value for 3 months after 7.1(±0.2)). *Values found by us after analysis of the primary data which are different from the values published in the article.However, the new values did not reach statistically significant differences (p=0.687).
31
Risk of bias of individual studies
Figure 2 displays the risk of bias classification for randomized studies, while
Figure 3 and Figure 4 depicts such classification for non-randomized studies. A detailed risk of bias assessment, together with support for the judgment and algorithm/tables used to complete the overall risk bias of all studies are presented in Appendix II for each outcome of each study.
32 Figure 3. Risk of bias for non-randomized studies for saliva composition and caries (by robvis tool119)
33 Figure 4. Risk of bias for non-randomized studies for cariogenic bacteria counts ( by robvis too119)
The multicenter randomized trial carried out by Isaacson et al (2020)67 was overall
classified as having “some concerns” on its risk of bias, particularly due to concerns in outcome measurement. One of the posters115 based on this RCT67 stated that
participants were examined by a dentist but the author did not explain how the oral examination was performed, how carious lesions were diagnosed, or what was considered a carious lesion. Therefore, doubts remain on whether incipient lesions could have been unreported on account of the adopted diagnosis methods. The protocol from
34 this study mentions that four radiographic bitewings were to be taken and that Dental Adverse Event (DAE) criteria were followed as defined in the protocol. It also mentions that subjects or their caregivers also completed a dental questionnaire at screening.
However, neither the DAE criteria nor the Dental Screening Questionnaire were available in the publicly available trial protocol. In addition, no information was available about how dentists were trained and calibrated - this is a relevant aspect as caries were assessed by different dentists. The authors were contacted to provide additional information, but our questions were not fully answered.
In addition, according to the latest publication67 of the Mysticol study, dental
examinations were performed by independent dentists and, which could indicate that they were blind to the intervention received by patients, but that information was not clearly mentioned by authors.
The overall risk of bias of the RCT published by Wu et al. (2011)41 was judged to
be of “some concerns” in relation to oral pH. While the bias assessment tool had four of the five domains classified as “low risk of bias”, the fifth domain - bias in the selection of the reported results - was classified as “some concerns” because we have not found any publicly available research protocol or pre-specified analysis plan (and the corresponding author did not respond to contacts made). Regarding both S. mutans and Lactobacilii counts, the overall risk of bias was considered high for both outcomes, not only because no protocol was found, but also because this study did not provide data of cariogenic bacteria counts (the only given information concerns the fact that there were not statistically significant differences between compared groups). Of note, although the responses led, through the use of the algorithm provided by the Cochrane tool110, to a
“low risk of bias” classification in the domain related to the randomization process, we believe that it should have been classified as "some concerns" or “unclear” due to the fact that the publication does not report how randomization was carried out.
As for the non-randomized studies, the study by Moller et al (2015)101 was
classified as having a critical risk of bias. Of note, there was a critical risk of bias on the possibility of confounding the effect of the intervention since extra oral exams and oral hygiene instructions were performed. Another important issue of concern refers to the absence of a control group in this clinical trial and the fact that it is not clear about which population (intention to treat or per protocol) the authors refer to when they state that no participant has developed caries, raising suspicions if caries lesions could have been underreported if it considers only to the per protocol population. In this case, some data may have been lost in relation to participants who dropped-out101. Failure in outcome
definition has also contributed for this study to be classified as having a critical risk of bias.
35 The 2017 study by Tiigimäe-Saar et al55 was judged to have an overall critical risk
of bias for all outcomes. Not only there was no control group, but also there was serious possibility for confounding. In fact, participants were very heterogeneous regarding their underlying diseases and medication use, with no approach described for adjustment or control of confounding variables. In addition, outcome measurement involved subjective judgments by the evaluator, which could have been influenced by knowledge on the received intervention - for oral pH and buffer capacity, results were determined by the difference in color on a scale provided by the manufacturer salivary tests. For measuring both S. mutans and Lactobacilli counts, results were determined by differences in the number of adherent colonies on the slide and compared against a scale provided by salivary tests manufacturer.
The 2018 study by Tiigimäe-Saar et al (2018)94, was classified as having a
serious risk of bias. Although the authors presented results for two control groups, those were not considered in our analyses, as they did not have hypersalivation (with one of the control groups even consisting of healthy individuals). There is also a risk of selection bias, more precisely indication bias. On the other hand, there is a serious risk of bias regarding missing data because the only drop-out was from the active group and he did not perform the post-intervention salivary exams. As the active group was composed of only 12 participants, we observed a dropout rate of 8.3%. Within the scope of the biases related to outcomes measurement, we observe the same problem as that of the 2017 study55, namely the subjective measurement method for all outcomes. Finally, there is
possibility of selective reporting, as, one month after applying the toxin, researchers only performed salivary tests in the active group, despite the protocol saying that the tests would be carried out in all groups.