2.1 Type of Study

Systematic review following the PRISMA 2020 guidelines [21]. The protocol was previously registered in PROSPERO (International prospective register of systematic reviews) (CRD42023435653) on June 25, 2023.

2.2 Search Strategy and Selection Protocol

We used the thesaurus MeSH (Medical subject headings) to identify synonyms for the terms lacrimal gland and lymphocytic infiltration. To increase the search sensitivity, we also performed a Pearl Harvest, in which the following additional terms were identified: Dry Eye Disease, Dry Eye Syndrome, Keratoconjunctivitis Sicca, Sicca Syndrome, Sjögren Syndrome, Xerophthalmia and Dacryoadenitis. Healthy controls are commonly included as a reference in the studies of these conditions. After selection of the search terms, we applied the search in PubMed, Science Direct and Google Scholar as specified in Table 1. Also, the reference section of the selected articles was reviewed to identify other relevant articles for the study. Duplicates were eliminated manually using filters in an Excel database.

2.3 Screening and Eligibility Criteria

The following inclusion criteria were applied: publications with search terms in the title, abstract or keywords, studies in English, Spanish or Portuguese, articles containing original results, studies in humans, studies with histologic evaluation of the LG and those that quantified lymphocytic infiltration in the healthy lacrimal gland. No time restrictions were applied to the search. Unavailable articles and studies that did not include the evaluation of the lymphocytic infiltration in the LG were excluded.

2.4 Data Extraction, Reproducibility, and Evaluation of the Methodological Quality

The following information was extracted from the selected articles: Title, authors, year, journal and country of publication, type of study, type of LG included (either main or accessory), number of healthy LGs included, number of LGs analyzed per individual, age and sex of the individuals included, type of sample (from either cadavers or biopsies), criteria used to define healthy individuals, methodologies, molecular targets and criteria used to quantify the infiltration and finally, results regarding the type of infiltrates, the location and the association with age and sex.

The search and selection of studies, and the data extraction were carried out by two researchers independently and discrepancies were resolved by consensus.

The methodological rigor of the selected studies was evaluated using the criteria of methods section of the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) guide [22].

2.5 Information Analysis

Qualitative synthesis of the extracted variables was carried out. From the studies that reported the degree of infiltration according to the age group, we dichotomized the infiltration as mild or moderate-severe and the age as younger or older of 60 years of age. A meta-analysis for the odds ratios was conducted evaluating homogeneity with Q statistic and I²; the sensitivity analysis was carried out with the influence graph, the evaluation of possible publication biases with the Begg statistic, and the general result was reported with the Forest Plot. The analyzes were carried out in SPSS 29.0, with a p < 0.05 considered significant.

3.1 Selection of the Studies and Analysis of the Methodological Rigor

The search terms in the databases and in the manual search resulted in a list of 1496 and 24 articles for screening, respectively. From them, a total of 20 articles were included (Figure 1). The only article excluded in “others” was the article by Belfort et al., 1980 [23] because of the use of an obsolete methodology.

Because the 20 selected studies were classified as observational, the nine criteria from the methods section of the STROBE guidelines were applied to evaluate methodological rigor [22]. The results from this analysis are shown in Figure 2. Only Sato EA et al., 2010 specified the study design [24]. All the articles provided information about the study settings and data collection. From the 14 studies using cadaveric LG, only seven specified their collection by autopsy [4, 16, 18, 25-28], others were collected either from an eye bank [29] or from LG excess tissues [30]. Five of the articles do not specify cadaver's origin [11, 19, 20, 31, 32]. From the eight papers that used LG from biopsies of living subjects, six specified their obtention during LG unrelated ocular surgeries [11, 17, 27, 33-35] and only in two studies, the biopsies were performed exclusively for the related study [24, 36]. We found information about the place and relevant dates of data collection only in eight [16, 18, 19, 25, 26, 30, 32, 36] and two articles [18, 33], respectively.

Two studies did not provide information about the eligibility criteria [28, 30]. In general, the remaining studies classified the healthy individual by the absence of autoimmune, systemic, connective tissue-related or ocular diseases. Wieczorek R. et al. and Franklin RM et al. included patients with eye tumors, but not involving histologically the LG [17, 34].

Regarding the definition of variables, all studies specified how lymphocytic infiltration was evaluated. Only Roen JL et al. quantify qualitatively the infiltration (mild, moderate) [28]. All the studies specified the data sources and measurements, however, only 11 (55%) applied descriptive statistics or statistical methods in the analysis. Three studies (10%) describe the efforts to address potential sources of bias [11, 25, 26]. Lastly, none of the studies explain how the study size was calculated, presumably because the LG were included upon availability.

3.2 Description of the Selected Studies

Table 2 specifies the characteristics of the apparently healthy individuals used in each study. The 20 selected studies included 774 healthy individuals. Fourteen studies were classified as cross-sectional and six were case-control studies (Table 3).

Table 4 presents the age group's distribution in all the individuals with healthy LG in the eleven studies with available information. This represents 460/774 (59.4%) from the total population. We found a minor representation (12.6%) of individuals younger than 40 years whereas more than a quarter of the population belonged to the age group of 61-70 years. It was not possible to extract the age group's distribution of the healthy individuals in nine of the selected studies (Table 4). Some studies did not provide the age of the participants [33, 34, 36]. The other studies presented the age data by using descriptive age statistics as follows: Waterhouse J.P., 1963 [16] 226 controls with an age range 0–75; Williamson J. et al., 1973 [31] 10 controls with an age range 32–76 and a mean of 64; Brandzaeg P. et al., 1979 [35] 13 controls with an age range 2–65; Pepose J.S. et al., 1990 [32] 7 controls with an age range 25–65; Sato EA et al., 2010 [24] 11 controls with a mean of 48.6 ± 14.9; Damato B.C. et al., 1984 [11] 97 controls with an age range 7–93 and a mean ± SD 62 ± 17.

Table 5 specifies the sex per age group of the individuals in the six studies with available information (303/774, 39.1% of the population included). The female:male distribution in these studies was 38.9 versus 61.1% with a greater representation of males in the age group from 41 to 80+ years.

3.3 Methods of Evaluating the Lymphocytic Infiltration of the LG

In the 20 selected studies, we found three methodologies to evaluate lymphocytic infiltration: hematoxylin-eosin staining (H&E), immunohistochemistry (IHC), and indirect immunofluorescence (IF). Twelve studies evaluated lymphocytic infiltration by H&E [4, 11, 16-18, 20, 24-26, 28, 31, 36]. However, the quantification was very heterogeneous. Four studies [16, 18, 25, 26] used the term “focus” for quantification, defined as an aggregate of 50 or more cells. However, Wieczorek R. et al. [17] defined focus as an aggregate of 30 cells or more. Only Segerberg-Konttinen M. used the term Focus score, defined as the number of foci in a 4-mm² area of gland tissue [26]. Waterhouse JP quantified foci/section in 1-cm² of lacrimal tissue [16]. Murray S.B. et al [20] neither defined the “focal” aggregates nor the atypical “focal” inflammatory cell infiltrate. On the other hand, Roen J.L. et al., Damato BE et al and Segerberg-Konttinen M. measured infiltration through dichotomous variables whereas Sato E.A. et al. calculated the inflammatory cell density per mm² [11, 24, 26, 28]. Only Williamson J. et al. classified the lymphocytic infiltration in stages and also took into account the structural changes in the gland [31]. Two studies quantified plasma cells in the LG by H&E [27, 36]. We illustrated in Table 6 the great variability in the studies that measure lymphocytic infiltration severity by grades or stages.

As shown in Tables 7 and 8, we also found great variability in the quantification of lymphocytic or plasma infiltration in the LG, respectively, using the different methodologies.

3.4 Quantification of the Lymphocytic Infiltration in the Lg from Healthy Individuals

Putting together the information regarding the main lymphocyte subpopulations in the healthy human LG may help to define criteria for the normality of immune cell infiltration as well as to assess the role of the immune system in DED and related conditions. Moreover, these studies may help to corroborate the information obtained from animal models. From the studies with available data, Sato EA et al, 2010 [24], reported the inflammatory cell density of 377 ± 152 cells/mm² in nine LG, a value significantly lower to that obtained from SS patients. The three IHC studies that quantified general lymphocyte subpopulations in the gland used different measurements (Table 7) (Figure 3) [17, 19, 30]. A consistent finding in these studies was the greater proportion of suppressor or cytotoxic (CD8 + ) as compared with helper T cells (CD4 + ) in the LG. This finding is also confirmed by Pepose JJ et al [32] who estimated the average of CD4 + T cells in 5-25% of total number of infiltrating cells in the seven control glands analyzed whereas CD8 + T cells represented 25-50%. Also, Dua HS et al, 1994 [29] described that CD8+ are twice as many as CD4 + T cells in the LG. Gudmundsson OG et al [19] compared the T cell proportion by sex and observed a slight but not significant increase CD4 + T cell absolute counts in females. The CD4:CD8 ratio in the LG from females and males reported in the later study was 0.74 and 0.57, respectively [19]. However, Wieczorek et al reported a CD4:CD8 ratio of 0.56 in the seven female LG analyzed [17].

Together with the CD4+ and CD8 + T cell values, Wieczorek et al calculated the absolute counts of B and plasma cells in the seven female LG [17]. Plasma cells were the most abundant lymphocytes in the gland (158 cells/mm³ or 53.9% of the total average mononuclear cells, (Figure 3C), a finding only qualitatively confirmed by Sacks EH et al [33] in an accessory LG of Krause from a superior forniceal specimen. These authors reported abundant plasma cells (stained with the OKT10 antibody, that target CD38) in the interstitium, between the acini of the glandular tissue, admixed with helper and suppressor T cells. B cells represented only a small proportion of the lymphocytes present in the human LG according to Wieczorek et al [17] (Figure 3C). However, Sacks EH et al [33] did not detect B cells in the accessory LG analyzed, by using three different monoclonal antibodies.

Figure 4 shows the distribution of the Ig-expressing plasma cells in the healthy human LG in the studies with available information. IgA-expressing plasma cells are the most abundant, followed by those expressing IgD, IgG, IgM and IgE [11, 34, 35]. Allansmith et al [27] reported the IF plasma cell patterns per 100 acini in accessory LGs, however, their results lacked sensitivity. Wong AJ et al, 2014 [36] reported IgG4-expressing plasma cell counts of less than 10 per high power field in seven controls. Other interesting data was presented by Allansmith MR et al [27], who observed greater plasma cell counts in the main LG as compared to the accessory gland and conjunctiva, despite the fact that the latter is the tissue most exposed to the external environment. They also reported an average count of 188 IgA1/IgA2+ cells/section (range: 58-356) [27]. Another report from the same author [4] described the average percentages of IgA1- versus IgA2-expressing plasma cells in the LG (56 vs. 44%), with no difference between the 10 main and one accessory glands from controls compared to one from a SS patient. Wieczorek R et al, 1988 reported that kappa predominates over lambda light chains in a proportion of 2:1 [17].

3.5 Distribution of the Lymphocytic Infiltration in the LG from Healthy Individuals

In contrast to the information about the lymphocytic infiltrate counts and percentages in the LG from the selected studies, the evidence regarding the distribution of the different lymphocyte subpopulations was very consistent. Pepose JS et al observed T helper cells scattered within the interstitium, predominantly at the periphery of the lymphocytic foci. Only a few cytotoxic-suppressor T cells were scattered throughout lymphoid aggregates, however, this was the predominant phenotype in the interstitium, far from the lymphoid aggregates and interspersed between acinar cells. B cells were found in the center of primary follicles and in occasional secondary follicles [32]. A similar distribution had previously been reported by Wieczorek R et al who also compared CD4+ versus CD8 + T cell counts specifically in the lymphoid aggregates and primary follicles from five of seven glands. The CD4:CD8 average ratio in this location was 2.2. They also reported that only 16% of the CD4 + T cells preferred glands or ducts to the interstitium [17]. Dua HS et al also reported suppressor CD8 + T cells in the proximity of the acini and close to the epithelial lining of the ducts [29]. Gudmundsson OG et al described small to medium sized T-lymphocytes in the interacinar tissue, often adjacent to an acinus or close to a collecting duct [19].

Regarding Ig-expressing plasma cells, Murray SB et al observed them scattered around the acini and throughout the fibrous septa [20]. Franklin RM et al reported increased numbers of IgA-expressing plasma cells in the interstitium [34], a finding validated by Wieczorek R et al [17] and also by Sacks EH et al [33] in an accessory LG.

Obata H et al [37] calculated the percentages of lymphocytic foci and periductal lymphocytic infiltration as evaluated by H&E, in the palpebral vs the orbital LG lobes. In LG with two or more lymphocytic foci per section, the orbital lobe exhibited significantly more lymphocytic foci.

3.6 The Association of the Lymphocytic Infiltration with Age and Sex

The association of the lymphocytic infiltration with age and sex specified in the selected articles is summarized in Table 9. There was no association between the Ig-producing plasma cell counts or distribution, with age and sex in three studies [4, 11, 27]. By using H&E, Waterhouse JP and Nasu M et al observed significantly more lymphocytic infiltration in women older than 45 years or individuals older than 40 years, respectively [16, 18]. However, two other studies did not report this association [19, 20].

Therefore, we collected the available information and dichotomized the data about lymphocytic quantification into null/mild or moderate/severe per group age, to perform a meta-analysis. Only in three of the selected studies it was possible to retrieve this information [18, 26, 28] from a total of 249 individuals (32% of the LG from the selected articles). The dichotomization was performed as follows: In Nasu M et al, grade zero or no infiltration was classified as null/mild and the other grades as moderate/severe [18]. In Roen JL et al, no infiltration was classified as null/mild and the presence of inflammatory cells (in 15 of the studied LG) was classified as moderate/severe [28]. In Segerberg-Konttinen we considered a focus score of less/equal or greater as 1 as either null/mild or moderate/severe infiltration, respectively [26]. Waterhouse JP and Murray SB et al only presented the data about the lymphocytic infiltration in the LG by age group in figures and therefore, were excluded from this meta-analysis [16, 20]. The analyses of null/mild versus moderate/severe lymphocyte infiltration per age group was only possible between individuals either younger or older than 60 years and showed no statistical difference. We neither observed statistical heterogeneity nor a publication bias among the studies. Roen and Segerberg-Konttinen include patients with AR and DM (five and two, respectively) that were not possible to separate in the analyses [26, 28].

The combined results of the meta-analysis showed no statistical differences between the degree of infiltration and the two age groups given that the odds ratio combined (with a random effects model) was 1.5 (CI 95% 0.7; 3.2) (Figure 5).