Unstimulated whole salivary flow in Sjögren’s Syndrome: systematic literature review and meta-analysis

Sjögren’s Syndrome compromises the exocrine function, producing xerostomia and xerophthalmia. It can appear as an isolated condition or associated with other autoimmune diseases (polyautoimmunity). The Unstimulated Salivary Flow rate (UWSF) is used to quantify saliva production. There is no objective evidence to differentiate the values in patients with Sjögren’s versus healthy people or patients with non-Sjögren’s sicca. The objective of the present review was to evaluate the UWSF in patients with Sjögren’s syndrome in comparison to controls (healthy and non-Sjögren’s sicca patients). A systematic literature review was carried out (PRISMA guidelines). Analytical observational studies of cases and controls, cross-sectional studies, cohort studies and randomized clinical trials (including healthy controls) were considered. The Medline/OVID, Lilacs, Embase, and Cochrane/OVID databases were consulted. MeSH, DeCS, keywords, and Boolean operators were used. The meta-analysis (RevMan 5.2) was done through the random-effects model [mean difference (MD)]. Level and quality of evidence were evaluated by the Oxford Center Levels of Evidence and Joanna Brigs list respectively. Thirty-two articles were included (20 were case-control studies, 6 were cross-sectional, 2 prospective cohort, 2 retrospective cohort, and 2 studies were abstracts) and 28 were meta-analyzed. The unstimulated whole salivary flow rate in the Sjögren’s group was lower than in controls (healthy and patients with non-Sjögren Sicca syndrome) (MD-0.18 ml/min; 95% CI, − 0.24 to − 0.13; chi2-P-value < 0.00001). Heterogeneity was 97% and there was publication bias (funnel plot). The level of evidence was mostly 3 or 4. The quality of evidence was met (97% of items valued). For the first time, the unstimulated whole salivary flow rate is found to be lower in patients with Sjögren’s syndrome compared to controls (healthy and non-SS sicca) through a meta-analysis. PROSPERO CRD42020211325.

SS is among the most frequent systemic autoimmune rheumatic diseases. The prevalence is around 0.3 to 1 case per 1000 persons [3], and it has an annual incidence of 6.92 cases per 100,000 people [4]. It is a frequent condition in women with a 9:1 female/male ratio and a high frequency of diagnosis at the age of 50 [3].
The physio-pathological mechanisms of SS are unknown. However, the epithelial destruction of the exocrine gland by an abnormal T (especially Th1 and Th17) and B cell response to specific auto-antigens such as Ro/ SSA and La/SSB (autoimmune epithelitis) is described. The etiology for the development of this disease is multifactorial and includes genetic, epigenetic, hormonal, and environmental factors [5].
SS is classified according to American College of Rheumatology/European League against Rheumatism (ACR/EULAR) criteria. It includes the presence of ocular (objective tests) or oral gland hypofunction (by means of sialometry), specific autoantibodies, and focal lymphocytic infiltrates in the minor salivary gland. The ACR/ EULAR criteria have to be applied to patients with suspected SS due to dry eye and dry mouth symptoms (at least one positive response from five established dryness questions), salivary gland enlargement, or extraglandular compromise (at least one item from ESSDAI (EULAR Sjögren's syndrome disease activity index)) [6].
Up to 80% of patients suffer dry eye and dry mouth symptoms, the key manifestations of SS. The clinical picture is also associated with systemic (e.g. asthenia and articular pain) or extraglandular manifestations in 40% [3].
There are procedures that objectively measure saliva production. The sample can be taken from an isolated salivary gland or several that are combined during rest [Unstimulated Whole Salivary Flow (UWSF)] or with stimulation [Stimulated Whole Salivary Flow (SWSF)] [7]. The measurement of UWSF is a quick, easy, and non-invasive test, and it is an item in the classification criteria for SS. Recently, the EULAR developed recommendations for the management of Sjögren's syndrome with topical and systemic therapies [8]. They emphasized that the therapeutic approach to oral dryness should be driven by the baseline measurement of salivary glandular function (UWSF), and not by the patient's subjective feelings since environmental and personal stress factors may influence the subjective feeling of dryness.
The main concern here is the production of saliva measured by the UWSF. Several studies, including systematic literature reviews and meta-analyses have evaluated saliva production in patients with xerostomia in different non-SS autoimmune diseases such as systemic lupus erythematosus, type I diabetes mellitus [9], autoimmune thyroid diseases, oral lichen planus [10,11], systemic sclerosis [12], juvenile idiopathic arthritis [13]. All of these have shown a lower rate of saliva production compared to healthy controls. Nevertheless, we were unable to find saliva production results for patients with SS as a primary disease in our search.
In the context of SS, a value of ≤0.1 ml/min is included in the ACR/EULAR SS criteria [14]. Nevertheless, this cut-off point went through a transition over a period of years, and it is included in or excluded from the different proposed international criteria [6,[14][15][16][17][18][19][20]. However, the justification for these changes is not clear in the literature even though it is a diagnostic criterion with the same weight as eye tests (Schirmer and ocular staining score) for an SS diagnosis. The established values in these criteria differentiated salivary gland hypofunction only in patients with sicca symptoms versus healthy controls, but not in patients with SS.
Therefore, in this systematic literature review and meta-analysis, the intention is to evaluate the saliva production through the UWSF in SS patients compared to non-SS patients with sicca symptoms and healthy controls in order to show objective differences between the diverse study groups.

Inclusion criteria Study design
Analytical observational studies of cases and controls, cross-sectional studies, cohort studies and randomized clinical trials (including healthy controls).

Population
Patients 18 years of age or older, with sicca symptoms and SS diagnosis based on established criteria: ACR/ EULAR, AECG (American-European Consensus Group), and Fox et al. [14,18,22].

Intervention
UWSF measurement by any method of collection.

Control group
Patients with non-SS sicca symptoms and/or healthy subjects.

Exclusion criteria
Studies that compare with other autoimmune diseases. Pregnant women.

Data extraction and quality assessment
Two reviewers (NA and CG) independently screened titles and abstracts to determine the initial eligibility. After the screening phase, the same two reviewers evaluated the selected abstracts and full text independently (software Mendeley Desktop v1.19.2) to evaluate their inclusion based on the established criteria. Another two reviewers (ARV and MAM) resolved disagreements by discussing them until they reached a final consensus.
The level of evidence and methodological quality assessment was done by two independent reviewers (NA and ARV) using the Oxford Centre for Evidence-Based Medicine 2011 Levels of Evidence Question (OCEBM) [23] and Joanna Briggs Checklists [24].
Two independent reviewers (MAM and JPA) extracted the data using a Microsoft Excel® (v. 2013) database. The UWSF measures were unified to ml/min, and the values reported in the subgroups were averaged. The central tendencies and dispersion measures (mean and standard deviation) were imputed. The expert reviewer (ARV) independently verified the abovementioned values. Two authors (JPA and MAM) registered the extracted data in the Review Manager Software 5.2® and the two authors' inconsistencies were corrected.
The following data were taken out: First author name, study year, study design, objective of the study, country, number of cases with SS (primary SS and SS with polyautoimmunity), number of controls (non-SS sicca symptoms and/or healthy subjects), number of women, age, disease duration, UWSF, stimulated salivary flow rate (SSF), comorbidities, and pharmacological therapy in cases and controls.

Data analysis
Statistical analysis was done with Review Manager Soft-ware® (RevMan 5.2). After that, data from the articles chosen were merged using the random-effects model (DerSimonian and Laird) due to the studies' high heterogeneity (sample size, saliva collection technique, and classification criteria for SS) to establish a global summary measure (overall effect).
The meta-analysis measurement considered the possible effects of the exposure as an average of ranges. The final effect size was presented as a mean difference with a confidence interval and the respective P-value. The forest plot was used to show the weight of each study.
Heterogeneity was assessed using the Chi 2 statistic, and the I 2 (Higgins). I 2 values of 25, 50, and 75% corresponding to low, medium, and high levels of heterogeneity respectively were considered. In addition, substantial heterogeneity was established if the value of the Chi 2 test was less than one. To evaluate the robustness of results and to identify the potential sources of heterogeneity, a sensitivity analysis and a subgroup analysis were done. Publication bias was graphically assessed by visual inspection of the funnel plot and confirmed using the Egger's regression test.

Search results
The first search identified 569 articles. Two reviewers evaluated the studies by title and abstract and chose 133 without duplicates. The second round was done based on full text and 32 articles were selected. There was a total of 28 studies included in the quantitative synthesis. Figure 1 shows the PRISMA flow-diagram. Causes of exclusion are detailed in Table 2.
Of the articles included by full text, 20 were casecontrol studies, 6 were cross-sectional, 2 prospective cohort, 2 retrospective cohort, and 2 studies were abstracts and there was not enough information to establish the type of study and quality of evidence. The level and quality of evidence are shown in Table 3. The level of evidence for most of the articles was 3 or 4 using OCEBM guidelines and the quality of evidence was around 97% for all articles.
The 32 articles included correspond to approximately 3262 patients. Of these, 46.8% (1528) were cases, and 53.2% (1734) were controls. The age range of cases was 39 to 63 years old while the age range was 24 to 69 for controls. The female sex was the most frequent in both groups. With respect to patients with SS, 66% (1010) had pSS, around 25.3% (386) had SS, and 8.6% (132) suffered from polyautoimmunity. Furthermore, patients with non-SS sicca came to 54.5% (946), and healthy people were 45.5% (788) of the group of controls. In total, 28 articles were meta-analyzed to evaluate the differences in UWSF between patients with SS and controls (healthy subjects and patients with non-SS sicca). The UWSF in the remaining four articles could not be analyzed because the authors did not report absolute values. The total number of people in these 28 articles was 2759 patients and of these, 49.1% (1356) were cases. Of these, 63.8% (866) had pSS, 26.5% (359) had SS, and 9.7% (131) had polyautoimmunity. Around 93% (725) were female. Of the total, 50.9% (1403) were controls, and of the controls, 43.8% (615) had non-SS sicca symptoms and 56.2% (788) were healthy subjects. Female sex was also predominant and accounted for about 83.74% (567).
Of the studies included in the meta-analysis, 13 were based on standardized methods of which the most widely used is that of Navazesh et al. [25], followed by Vitali et al. [26], and Sreebny et al. [27]. The method used by Navazesh et al. [25], requires fasting, no smoking, no chewing gum, nor having any oral cavity procedures done for at least 1 h before the exam. It also requires 5 min of rest prior to the test during which orofacial movements must be minimized. Then, residual saliva must be swallowed before starting the procedure, and saliva must be expelled every minute for a period of 15 min during it.

Meta-analysis results
The pooled analysis demonstrated a lower UWSF in patients with SS [Mean difference (MD) -0.18 ml/min; 95% CI, − 0.24 to − 0.13; chi 2 -P-value < 0.00001] compared to controls (healthy subjects and patients with non-SS sicca) under a random-effects model. The I 2 statistic, which estimates the total variability in the effect size, was calculated to be 97% (Fig. 2) and could be attributed to heterogeneity among the true effects.
A subgroup analysis (heterogeneity of 98%) of patients with pSS reported that they had a lower UWSF compared to controls (healthy subjects and patients with non-SS sicca) (MD − 0.24 ml/min; 95% CI, − 0.32 to − 0.16; chi 2 -P-Value < 0.00001). The comparison of pSS versus non-SS sicca disclosed a difference in the UWSF (MD-0.10 ml/ min; 95% CI, − 0.13 to − 0.06; chi 2 -P-value< 0.00001 with a heterogeneity of 82%). As for the subgroup analysis (heterogeneity of 99%) of patients with pSS versus healthy people, a difference in UWSF was observed (MD-0.43 ml/ min; 95% CI − 0.49 to − 0.27; chi 2 -P-value < 0.00001). In a subgroup analysis (heterogeneity of 97%) by type of study design, the results were different but consistent with the pooled results from all subgroups with the exception of the cross-sectional study group in which there were no significant differences (Fig. 3).

Discussion
Here, for the first time (to our knowledge), a metaanalysis shows that the unstimulated whole salivary flow rate is lower in patients with Sjögren's syndrome compared to controls (healthy and non-SS sicca patients). A mean difference value of − 0.18 ml/min (95% CI,-0.24 to-0.13, P-value < 0.00001) is disclosed.
The above result is important given the role that saliva plays in oral health. For example, saliva provides the teeth with protection and lubrication and fosters their remineralization. It also cooperates with the transportation of digestive enzymes. It has antimicrobial properties and plays a part in chewing and swallowing. It is estimated that the saliva production rate is 0.5 to 1.5Lt/day in healthy adults [7,28]. The glands responsible for saliva production are the major salivary glands (parotid, sublingual, and submandibular) and account for 90% of it. The remaining 10% of saliva production is attributed to the minor salivary glands that are diffusely distributed in the oral mucosa [7,29]. Diminished production of the salivary flow rate implies the hypo-function of salivary glands. Therefore, it is essential to find out what the main causes of the compromised flow rate are. The adverse effects of medication, therapeutic irradiation, and SS are common causes of decreased salivary flow. The ACR/EULAR criteria are used for the SS diagnosis. They include the UWSF rate, and if the salivary production is ≤ 0.1 ml/ min or ≤ 1.5 ml in 15 min, the criteria assign one point to the final calculation. Unfortunately, in the majority of the cases, this test is only applied when the patient does not have dry eye by objective measures (Schirmer test or ocular staining score) [30]. In general terms, there is a consensus in the literature on the UWSF cutoff point that demonstrates glandular hypofunction when its value is ≤ 0.1 ml / min. However, it has been suggested that values between 0.1 to 0.2 ml/min could be the lower threshold, and that values > 0.2 ml/min are found in healthy people [31].
The value of ≤0.1 ml/min includes all the people with dry eye/mouth symptoms resulting from causes that cannot be differentiated from SS. Traditionally, this cutoff point of UWSF is used for the SS diagnosis. However, the origin of this value is not clear in the literature review (Table 4). In 1975, Daniels et al. said that there were no established specific criteria for determining the presence of the oral component of SS, but since then, the abovementioned cut-off point has been used in different sets of criteria for patients with suspected SS with no solid scientific support [15]. Thus, this Systematic Literature Review (SLR) and meta-analysis were developed to evaluate the difference using an objective measure. In the present study, the difference in the value of the UWSF for controls and the one for SS was found to be − 0.18 ml/min. Note that out of the 1403 controls included, a considerable group were healthy subjects (n= 788). That is why, when a subgroup analysis is done to evaluate the difference in UWSF between pSS and healthy people, it shows a higher than expected difference, − 0.43 ml/min. Furthermore, the controls with non-SS sicca in the present results include subjects with sialosis, fibromyalgia, radiotherapy, burn mouth syndrome, diabetes, arterial hypertension, sodium retention syndrome, dyslipidemia, positive serology for hepatitis C, hyperthyroidism, hypothyroidism, osteoarthritis, and neoplasia. A minor difference in the saliva production (− 0.10 ml/ min) was found in the subgroup analysis of patients with pSS and non-SS sicca as was expected. In a 2016 systematic literature review that evaluated the level of xerostomia, hypo-salivation, and salivary flow rate in patients with diabetes, the UWSF rate was found to be 0.16 to 0.55 ml/min in the group of patients and 0.26 to 0.75  ml/min in the control group. This is similar to the present results even though this population corresponds to non-autoimmune diabetes (Type 2) [32].
Another meta-analysis from 2018 involved subjects with autoimmune thyroid diseases and oral lichen planus. SS was eliminated from the study, and it included 130 patients in the end. The authors analyzed UWSF values and found a diminished production of UWSF in the group of patients compared to controls (− 1.09; − 1. 49, − 0.7; P-value < 0.001) [10]. It reveals the lack of studies that include patients with other autoimmune diseases. However, it shows evidence of a greater difference in the mean value of UWSF in cases versus controls in comparison to our meta-analysis of values of UWSF of patients with SS versus controls.
The clinical usefulness and practical implications of the present results show that cases have − 0.18 ml/min of unstimulated salivary flow compared to controls. This implies that in 1 h, patients present a saliva production of 11 ml less than controls, directly impacting previously mentioned basic physiological functions. The present meta-analysis results could be involved in future studies to define new UWSF cut-off points that better discriminate between patients with SS vs. sicca non-SS by analyzing sensitivity, specificity, and through ROC curves.
In addition, this could guide new research focused on modifying the pre-stablished cut-off point for making treatment decisions. In fact, the recent EULAR recommendations for the management of Sjögren's syndrome [8] suggested a first evaluation considering the UWSF measurement, and if the value is < 0.1 ml/min, they proposed to classifying the patients according to SWSF values (> 0.7 ml/min: normal/mild, 0.7-0.1 ml/min: moderate and < 0.1 ml/min: severe dysfunction). If SWSF values showed severe dysfunction, saliva substitution should be considered the preferred therapeutic approach to alleviate symptoms. This would be needed given that in patients with no residual glandular function (severe glandular dysfunction), salivary glands cannot be stimulated, either by pharmacological or non-pharmacological interventions. The present meta-analysis results might guide new discussions regarding practical implications with respect to not only the diagnostic process, but also, treatment decisions given the limitations on developing the SWSF measurement. This is a complicated test in daily practice and not always available in all clinical settings as was mentioned by the EULAR panel in their recommendations. It might be feasible to discuss avoiding the SWSF test and using the UWSF instead given the objective saliva production measurement based on the present results.
The limitations of the present study include its heterogeneity through the I 2 test and the publication bias as evinced in the funnel plot due to the fact that it had a preponderance of studies with a small sample size (95% smaller than a n=100). Additionally, there are variations in the saliva collection technique and the sampling time in the different articles. Therefore, the results have to be taken with caution. Nonetheless, this exercise supports the need to unify criteria for clinical evaluation, monitor patients with sicca symptoms, and set up epidemiological studies with larger sample sizes and greater homogeneity in their comorbidities. Moreover, the study design included articles that were mostly cross-sectional and cohort studies with a three or four level of evidence. There were fewer articles that compared SS patients with diagnoses of other autoimmune diseases (polyautoimmunity) to pSS patients and, as a result, no difference was found. Therefore, this could be an important step for future research.
Unfortunately, the articles did not provide enough information to do a meta-regression. Furthermore, some of them were not clear with respect to whether or not the cases and controls were receiving pharmacological treatments or had comorbidities or other diseases that could have an influence on the values of the UWSF.

Conclusion
For the first time, the difference in the values of the UWSF rate in patients with SS compared to controls (non-SS sicca subjects and healthy people) was objectively demonstrated through a meta-analysis. The purpose behind this study is to challenge the different specialists in oral health and rheumatology regarding the importance of the use of UWSF, which has the same weight within the classification criteria [14] for SS as the Schirmer test or the ocular staining score test. This takes into account the fact that eye tests are done more frequently than the UWSF in real-life conditions in the context of SS diagnosis. The results of our research will make it possible to understand the objective assessment of cut-off points for UWSF in SS and in other causes of xerostomia.