Recent advances in age-related meibomian gland dysfunction (ARMGD)

Abstract

Meibomian glands (MGs), located within the tarsal plate of the eyelid, secrete meibum which is the lipid-rich secretion necessary for stabilizing the tear film and preventing tear evaporation. Changes in the quality and quantity of meibum produced causes MG dysfunction (MGD), the leading cause of evaporative dry eye disease (EDED). MGD is an underdiagnosed disease and it is estimated that, in the US, approximately 70 % of the population over 60 have MGD. Three forms of MGD occur based on their meibum secretion: hyposecretory, obstructive, and hypersecretory MGD. The pathophysiology of MGD remains poorly understood, however aging is the primary risk factor. With age, MGs undergo various age-related changes, including decreased acinar basal cell proliferation, hyperkeratinization, MG atrophy, and eventual MG drop-out, leading to age-related MGD (ARMGD). Additionally, studies have suggested that MGs can suffer inflammatory cell infiltration and changes innervation patterns with aging, which could also contribute towards ARMGD. This review focuses on how the aging process affects the MG, and more importantly, how age-related changes to the MG can lead to MG atrophy and MG drop-out, ultimately leading to ARMGD. This review also highlights the most recent developments in potential therapeutic interventions for ARMGD.

1. Introduction

Meibomian glands (MGs) are located within the tarsal plate of the eyelid and secrete meibum onto the ocular surface [1]. Meibum is a lipid-rich secretion that is necessary for stabilizing the tear film and for protecting the ocular surface from various hazardous factors. Changes in the quality and quantity of meibum cause an increase in tear film evaporation, leading to evaporative dry eye disease (EDED) affecting approximately 21 million individuals in the US, with these numbers increasing with the aging population [2]. The prevalence of dry eye disease (DED) worldwide ranges from 5 to 50 %, depending on different populations [2–4]. Studies have shown that DED has multiple etiologies, with Meibomian gland dysfunction (MGD) being the leading cause [3, 5]. In fact, clinical studies suggest ~85 % of all DED cases are caused by some form of MGD [6]. MGD is an underdiagnosed disease, and it is estimated that, in the US, approximately 70 % of the population over 60 have MGD [7]. Moreover, MGD is believed to be more prevalent in the Asian populations compared to Caucasians [8]. Symptoms associated with dry eye include burning sensation, itchy eyes, pain, fatigued and/or sore eyes, dryness sensation, red eyes, photophobia and blurred vision, all of which decrease productivity and reduce the quality of life [9]. With age, MGs undergo various age-related changes, including decreased acinar basal cell proliferation, MG atrophy, and eventual MG drop-out, leading to age-related MGD (ARMGD) [10,11]. Unfortunately, little is known about what causes these changes to the MG, making developing therapies difficult. Conventionally, ARMGD is believed to be caused by obstruction of the MG secretory duct, leading to stasis of the meibum within the duct, which, in turn, leads to backpressure within the gland triggering MG atrophy [12]. Hyperkeratinization at the opening of the collecting duct that occurs with aging has been regarded as a major cause of MG obstruction [13]. More recently it was proposed that a decrease in cell proliferation within the basal layer of the MG leads to reduced meibocyte differentiation, and, consequently, leads to MG atrophy and reduced meibum production, all culminating in ARMGD [14]. Other studies have speculated that, as with the lacrimal gland [15,16], the MG could suffer inflammatory cell infiltration with age that could contribute towards ARMGD [17,18]. More recently, studies have suggested that a loss in the number of MG progenitor cells over time could contribute towards ARMGD [14]. Additionally, some studies have speculated that changes in MG innervation patterns could also contribute towards ARMGD [19]. Finally, the eyelids also change drastically with aging, with a notable loss of elasticity and sagging. It has been speculated that changes in the biomechanical properties of the tarsal plate and overall eyelids could also contribute to ARMGD. Unfortunately, when studying naturally occurring ARMGD, it is impossible to separate each of these factors in order to understand their individual contributions to MG atrophy and eventual drop out (see Fig. 1).

Fig. 1.

Summary of the structural and functional alterations that occur with ARMGD, possible causes of ARMGD, and recently proposed therapeutic options for ARMGD.

Although the pathophysiology of ARMGD remains poorly understood, there are many well-defined risk factors. The primary risk factor for MGD reported to date is aging [1]. Other risk factors include hormonal imbalances, diet, eyelid defects, prolonged contact lens wear, excessive use of make-up, eyelid tattooing, and Demodex folliculorum infestation [20,21]. Certain systemic conditions can place individuals at a higher risk of developing MGD, such as, autoimmune diseases such as Sjögren’s syndrome, rosacea, lupus, psoriasis, and rheumatoid arthritis, and, Stevens-Johnson Syndrome (SJS), and hypertension [22,23]. Importantly, some lifestyle changes can help to slow the progression of MGD, for example dietary changes by increasing the ingestion of foods containing omega-3 fatty acids and practicing good eyelid hygiene [22].

Aging is a complex process that affects all cells and tissues of the human body. Traditionally, the hallmarks of aging were considered to be 1. genomic instability which results in DNA mutations, 2. telomere attrition leading to telomere shortening, 3) epigenetic alterations, 4. loss of proteostasis leading to the accumulation of misfolded proteins, 5) deregulated nutrient sensing, 6) mitochondrial dysfunction, 7) cellular senescence, 8) stem cell exhaustion, and 9) altered intercellular communication [24]. These changes that occur with aging all lead to an accumulation of damage to cells and tissues, which is aggravated by a decrease in the ability to repair damages that also occurs with aging. Therefore, it is not surprising that aging is a major risk factor in most diseases, including ARMGD. Herein, this review covers how the aging process affects the MG, and more importantly, how age-related changes to the MG cause MG atrophy and MG drop-out, ultimately leading to ARMGD.

2. Anatomy and morphology changes to the MG with aging

MGs are long linear structures that are serially arranged in both the upper and lower eyelids perpendicular to the eyelid margin [1]. Approximately 30–40 MGs are present in the upper eyelids, and 20 to 30 MGs in the lower eyelids [1]. MGs are formed of simple branched acinar glands that secrete meibum onto the ocular surface via a main central collecting duct [1]. Meibum is produced by holocrine secretion, and, therefore, as fully mature meibocytes reach the opening of the collecting duct, they rupture releasing the meibum and the remnants of the meibocytes into the collecting duct [1]. The blinking motion helps to move meibum through the collecting duct and in its eventual extrusion through the orifice onto the ocular surface [1]. Meibum secretion is believed to be passive, relying on the mechanical forces of the overlying orbicularis oculi muscle during the action of blinking [1]. Similarly, massaging of the eyelids can help to extrude meibum onto the ocular surface [25]. It is believed that meibum undergoes important chemical modifications while moving through the central collecting duct [1,26]. The blinking motion also helps to mix the meibum into the tear film where it forms the external lipid layer. MGs can be easily imaged in the clinic with meibographers, which are non-contact infrared photographic devices [27]. Currently, the cause, onset and mechanism of ARMGD are largely unknown. However, significant morphological and anatomical changes in lid margin and MGs are evident, as reported by various groups.

Gutgesell et al. (1982), conducted a histopathological study on MGs of seven human males in the age range of 58–83 years, all having clinical evidence of severe or moderately severe MGD. They reported abnormal features such as enlarged acini with compressed acinar cells towards the periphery of the acinus, indicating a more distal obstruction with stagnation of secretion. They also noted acini devoid of secretory contents, dilated acini with distinct lumina, dilated ductules and main duct, presence of small granulomas with multinucleated foreign-body giant cells with foreign-body reaction around acini, mild increase in chronic inflammatory cells around some acini and abnormal keratinization of ductal epithelium including sloughing of keratin, hyper-keratinization and narrowing of duct [13]. Another study by Hykin and Bron (1992) on 80 human subjects from different age groups ranging from <10, 11–20, 21–30, 31–40, 41–50, 51–60, 61–70, 70+ years (10 individuals, 5 males and 5 females, per group) explored age-related morphological changes in lid margin and MG anatomy [5]. The major changes outlined with aging included an increase in upper lid margin rounding, increase in lid vascularity, increase in lower lid telangiectasia, increase in cutaneous hyperkeratinization, squamous blepharitis, increase in prevalence of orifice pouting and narrowing, prevalence of orifice obliteration in upper lid, gradual increase in hyperemia, decrease in multiple rows of orifice in upper lid and decreased synthesis of meibum. No significant age-associated changes were observed in lid margin thickness, average lid margin cilia number (though they were less numerous in the very elderly), reduction or retroplacement of orifices, quality of meibum and the mucocutaneous junction [5]. Reneker et al. (2020), examined the histopathological features of MGs of aged human cadavers (63 and 64 years old males) and observed moderate to severe MG atrophy and significantly reduced cell proliferation as measured by Ki67 labelling index [28]. Recently, an artificial intelligence-based approach by Huang et al. (2023), using images of 60–89 year old human MGs obtained by noninvasive meibography (Keratograph 5 M), confirmed the previously established age associated changes in MG morphology parameters (e.g., area, density, number, height, width and tortuosity), lid margin parameters (telangiectasia, irregularity, thickening, plugging and lipid extrusion test score), meibum score and MG loss [29].

In mice, a dramatic decrease in total gland, ductal and lipid volume, absence of progenitor cells in atrophic glands, anterior migration of mucocutaneous junction and absence of ductal hyperkeratinization in aged mice (2 years old) was reported by Parfitt et al. (2013) [30]. Also, a reduction in acinar proliferation, presence of infiltrating plasmacytoid cells and a shift in the localization of PPARγ to nucleus was reported [18]. These structural changes were consistent with changes in the aged human MGs [17]. Overall, it is evident that MGs and eyelids undergo age-related morphological and anatomical alterations. Though the onset of ARMGD is still unknown, MG atrophy is already evident in humans by 25 years, and in mice by 6 months [33–36]. Both morphological and functional changes can be detected in early MGD [34,37]. Some of the early morphological changes that have been reported prior to MG atrophy are a decrease in the acinar size with a concomitant decrease in the number of cells per acini and accumulation of keratinized ductal epithelium within the MG [34]. Further investigation is needed to establish how early morphological changes to the MG that precede MG atrophy could be used in early diagnosis of MGD and ARMGD, which would enable early intervention and potential prevention.

The age-related changes that occur in the MGs of mice closely resemble those that occur in the humans MGs, and thus mice have proven to be a valuable animal model for ARMGD [18]. Specifically, MGs from mice presenting ARMGD resemble those of humans with ARMGD, in terms of altered cell cycle entry/proliferation, lipid synthesis and gland atrophy [18]. Additionally, certain breeds of dogs have been shown to have a high incidence of developing ARMGD. A case report of an aged (14-year-old) Cairn terrier dog presenting chronic inflammation and gland drop out, indicating ARMGD in dogs [31]. Another study by Vinas et al. (2003) confirmed age as a risk factor of MGD in dogs [32]. With increasing age, canine MGs show gland shortening and dropout along with ductal dilatation and gland tissue breakdown [33]. To the best of our knowledge, the existence of ARMGD in other species, including rats, rabbits and monkeys, remains to be established. Establishing ARMGD in higher order mammalian animal models would be invaluable to further our understanding on the pathophysiological mechanisms of ARMGD and for further developing therapeutic interventions for clinical trials. Unfortunately, none of the characterized animal models of ARMGD available to date recapitulate all of the clinical characteristics of human ARMGD. Despite the availability of various animal models for MGD and ARMGD in different species [34–36], there is still an unmet need for the characterization of additional animal models of ARMGD.

3. Changes in meibum composition with aging

The Meibomian glands attain functional maturity by 30th week in utero in humans and post-natal day 15 in mice [37] and start secreting the meibum. Meibum is composed of a rich mixture of lipids [38] along with some proteins [39]. The major classes of lipids present in human meibum include wax esters (WE), cholesteryl esters (CE), (O-acyl)-ω-hydroxy fatty acids (OAHFA) and their esters, free fatty acids (FFA), acylglycerols, cholesterol (Chl), diacylated diols (diacylated α,ω-diol), and a smaller amount of other polar and nonpolar lipids [40]. The protein component of meibum comprises of cytokeratins, cytokines, immunoglobulins, extracellular matrix proteins, proteoglycans, various enzymes, and others [39]. With increasing age, the human meibum undergoes both physical and chemical alterations which contribute to age-related MGD symptoms. Infrared spectroscopy based meibum analysis revealed that hydrocarbon chain order in meibum decreased from about 48 % of trans rotamers at birth to about 30 % trans rotamers at 85 years of age, indicating compositional changes in meibum with age. Also, the phase transition temperature of meibum is shown to decrease by 4 °C from about 31 °C at birth to about 27 °C at 90 years of age. These age-associated changes in human meibum have been attributed to compositional changes in meibum lipids [41]. Infrared spectroscopic studies have also revealed a decrease in lipid-lipid interactions with age, with human meibum lipid order decreasing from ~50 % at 3 years of age to ~27 % at 80 years of age. Since lipid order and protein content are related, the less protein content in aged human meibum seems to contribute to decreased lipid order with age [42]. Also, with age, the lipid carbonyls are shown to be hydrogen bonded or in a more polar environment [43]. The changes in the composition of meibum with age leads to changes in the physical parameters, such as color and consistency, with hypo-secretion [44] and hyper-viscosity [5] of meibum being associated with aging. Given the increase in meibum viscosity with ARMGD, different light therapies (e.g., low level light and intense pulsed light therapies) [45–50] have been proposed and practiced recently, as well as traditional warm compress therapies [51], for enhancing the flow of meibum, providing relief from EDED symptoms, as reviewed in Refs. [52–55].

Further changes in meibum lipid composition have also been identified by various other groups. Sullivan et al. (2006) revealed that aging is associated with significant changes in the neutral and polar lipid profiles of human meibum [56]. Suzuki et al. (2021), recently reported significantly increased polar lipids [cholesterol (Ch), (O-acyl)-ω-hydroxy fatty acid (OAHFA), and free fatty acid (FA)] in the meibum of elderly humans (aged 58.4 ± 7.5 years), with a significant decrease in non-polar lipids [e.g., cholesterol esters (ChEs)] and no change in WEs, using reverse-phase HPLC-MS/MS [57]. Also, the relative amount of CH3, C═C bonds and degree of oxidation increases with age, as revealed by nuclear magnetic resonance spectroscopy [58] and infrared spectroscopy [43]. The aldehyde-to-hydroperoxide ratio also increases with age, suggesting the progression of oxidation in the meibum of aged glands compared to younger ones [58]. Since meibum of aged individuals contains more CH3, C═C and aldehydes, relative to hydroperoxides, and are less ordered because of weaker lipid-lipid interactions and contain less protein; making tears of aged individuals less stable than infants/children [58]. On the other hand, Yeotikar et al. (2016) reported no significant change in five major lipid classes i.e., cholesterol esters (CE) with free cholesterol (FC), wax esters (WE), (O-acyl)-omega-hydroxy fatty acids (OAHFA), tri-acyl glycerides (TAG), and ceramides in the meibum of age individuals ranging from 25 to 66 years, using nano-electrospray ionization tandem mass spectrometry [59]. However, it is worthy to note that all subjects recruited in this study were asymptomatic with no eye disease or prominent dry eye complaints. In another study by Butovich and Suzuki (2021), age-associated changes in the meibum and meibogenesis were compared in human subjects of two age groups without any gender bias - young (29 ± 5 years, n = 21) and elderly with no MGD/DE (68 ± 7 years, n = 29), using liquid-chromatography mass spectrometry (LC-MS) and tan dem mass spectrometry (MS-MS) [60]. There were no statistically significant differences observed in the lipid esterification, elongation, and unsaturation patterns. Also, no difference with respect to length of carbon chain and degree of unsaturation was observed between these age groups. The relative abundance (%) data indicated that only changes in DiAD were statistically significant (p = 0.003), out of the major lipid classes studied (CHL + CE, WE, TAG, DiAD and Chl-OAHFA) [60].

The compositional heterogeneity of meibum in aged human subjects, like other meibum analysis studies, could be attributed to methods employed in meibum collection, incidental contamination (with skin/hair lipids, cosmetics), storage conditions (moisture, light, temperature, oxidation, plastic-containing container) and the analytical methods used [61,62]. However, in the past decade, availability of better sample collection & storage strategies and advanced analytical methods have helped in yielding contamination-free samples and sufficient amount of samples with reliable, sensitive and selective results. Thus, based on the current available literature, aging is associated (at least partially) with the changes in composition and/or amount of lipids present in the meibum, which in turn may contribute towards changes in its physical properties and result in symptoms of age-related Meibomian gland dysfunction (ARMGD). These changes are primarily, hyposecretion, hyperviscosity, significantly changed compositional ratios of non-polar lipids (decreased with age) and polar lipids (increased with age).

4. Correlation between MG inflammatory cell infiltration and MGD

Throughout life, tissues continuously suffer from a low-grade chronic oxidative insult, which increases with advanced age. This low-grade chronic oxidative insult results in low levels of activation of the immune system over time in an attempt to maintain homeostasis, which has recently been coined para-inflammation. Short term, para-inflammation is necessary for maintaining/restoring tissue integrity and functionality, however, over time para-inflammation can gradually cause the build-up of inflammatory cells and cause fibrosis. Thus, para-inflammation is an adaptive immune response where low levels of tissue stress trigger a low-level immune response throughout life leading to chronic inflammation [63]. Recently, chronic inflammation was added as a hallmark of aging [24]. Specifically, low-grade chronic inflammation occurs with aging, which was coined as “inflammaging” [64]. Inflammaging has been shown to have a key role in various age-related diseases, including cardiovascular diseases, neurodegeneration, and cancer [65–67]. Recently, it has been suggested that the MG may suffer inflammatory cell infiltration with age [17,18]. Many tissues in the body, such as the macula [63], retina [68], sebaceous gland [69], heart [70] and lungs [71], have also been shown to present gradual inflammatory cell infiltration with aging, which is associated with oxidative stress and loss of tissue function. Moreover, inflammaging has been shown to have a central role in the loss of function of the aging lacrimal gland [72]. Specifically, studies have shown that as humans and mice age, their LGs gradually suffer lymphocytic and plasma cell infiltration, which lead to acinar atrophy and fibrosis [15,16].

Studies have suggested that the MG presents a gradual increase in inflammatory cell infiltration with aging, however, the extent of inflammatory cell infiltration and whether it is associated with ARMGD remains to be established. An age-related increase in periglandular inflammatory infiltrates in the MGs of mice and humans has been previously identified and linked to ARMGD [17,18]. Moreover, the quality of meibum has been correlated with MG leukocyte infiltration, with the presence of CD45⁺ cells located within MG acini and/or ducts of MGs correlating with meibum production with reduced quality [18]. CD45⁺ cells, increased expression of inflammatory markers and increased fibrosis have also been identified in the periglandular area of the MG at 50 weeks of age in Cu, Zn-Superoxide Dismutase-1 (Sod1) knockout mice presenting increased inflammatory cell infiltration when compared to wt mice [73]. Interleukin-1ß (IL-1ß) is known to have proinflammatory properties and is increased in the tears of MGD and DED patients [74]. IL-1ß following Staphylococcus aureus infection leads to keratinization within the acini and ducts of the MG [75,76]. The effects of IL-1ß were also evaluated in MG epithelial cells, more specifically on the inhibition of the p38 MAPK pathway in adult and aging rats [77]. The p38 MAPK pathway has been suggested to regulate meibocyte differentiation, and the inhibition of the p38 MAPK signaling pathway was found to prevent the IL-1ß induced hyperkeratinization of MG ductal and acinar cells. Additionally, IL-1ß was found to increase MMP9 activity in isolated MG epithelial cells, which was speculated to potentially contribute to remodeling of the tarsal plate extracellular matrix during MGD [77]. Ibrahim et al. (2014) proposed an oxidative stress induced age-dependent MGD model using Cu, Zn-Superoxide Dismutase-1 (Sod1) knockout mice. These mice presented an age-related increase in MG periglandular inflammatory infiltrates, increased oxidative lipid and DNA damage, decreased MG glandular acinar density and an increased in periglandular fibrosis. Additionally, there was an increase in MG acinar cell apoptosis with acinar atrophy [78].

MGD and other MG pathologies can also lead to MG inflammation that is not related to ARMGD, named Meibomianitis [79]. Meibomianitis is often associated with chronic blepharitis and causes dilation of acini and ductules and leads excess accumulation of meibum at the eyelid margin [79]. Meibomianitis can also cause an imbalance/abundance of bacteria to grow in the inflamed glands [80–82]. Symptoms of Meibomianitis include swelling and redness of the eyelids, dry eye, slight blurring of vision due to excess oils in tears, and frequent styes [83]. Allergies, hormone changes, or skin conditions, such as rosacea, can cause Meibomianitis, in fact, two-thirds of patients with Meibomianitis have acne rosacea [82]. Meibomianitis has also been shown to cause keratoconjunctivitis, termed Meibomitis-related keratoconjunctivitis (MRKC) by Suzuki et al. [84], with redness and swelling around the eyelid margin and palpebral conjunctiva, and an abundance of P. acnes bacteria in meibum [84]. The cornea also presents with delayed-type hypersensitivity (DTH) to P. acnes and the severity and location of inflammation in the cornea depends on the degree of meibomianitis [83]. There is also subepithelial cellular infiltrates, superficial vascularization, and conjunctival hyperemia which is similar to phylctenular keratitis [83]. A clinical study showed that 87 % of women under 35 with phylctenular keratitis presented with MRKC [84]. Patients presenting with Meibmianitis also show changes in meibum composition where there are modifications of polar lipids [81]. Importantly, Meibomianitis can serve as a precursor of MGD by causing obstructive MGD [79]. Minocycline has been suggested as a treatment for Meibomianitis due to its ability to decrease diglycerides and free fatty acids through lipase inhibition or via a direct effect on the bacteria found at the eyelid margin [82]. Other common therapeutic strategies include topical and oral antibiotics such as azithromycin and doxycycline, respectively, and, steroids, and physical therapies such as MG massages, probing, LipiFlow, and intense pulsed light (IPL) along with meibum expression [1,79,85,86].

5. Proposed causes of MGD and ARMGD

a. Hyperkeratinization

MGD is a progressive disorder that typically presents with terminal duct blockage, with hyperkeratinization as the major cause of obstructive MGD [11,87]. MG hyperkeratinization is described as an excess accumulation of keratin formation [88] and plugging of the orifice that causes meibum stasis, cyctic dilation of the duct, and eventual gland atrophy and drop out [11]. It has been shown that keratinization can occur due to a higher presence of reactive oxygen species during MGD [89], causing the formation of more disulfide bonds during keratin crosslinking ([88]). Ong et al., 1991 studied the presence of keratin proteins through immunoblotting in healthy meibum and meibum of patients with MGD, and found that meibum from patients with MGD had a 10 % higher concentration of keratin proteins than healthy meibum [90]. An increase in the concentration of keratins within meibum has been shown to cause the lipids to become more rigid, increasing the surface pressure of meibum and decreasing its ability to stabilize the tear film, thereby contributing to EDED [88]. Studies have speculated that in ARMGD, acinar atrophy leading to MG drop out occurs as a consequence of aging, irrespective of hyperkeratinization [91]. Specifically, the expression pattern of cytokeratins (1, 5 and 6) in young and aged MGs revealed a lack of hyperkeratization in aging MGs, indicating MG obstruction is not the primary cause of ARMGD [30,91]. Thus, although hyperkeratinization is the most widely accepted cause of MGD, it is likely not a major cause of ARMGD. A study done using excess human eyelid tissue collected from patients which had undergone canthoplasty (ages 18 to 95) revealed that aging MGs showed decreased meibocyte differentiation and cell cycling through altered PPARγ signaling, resulting in MG atrophy and dropout, and, consequently, hyposecretory MGD [17]. The authors also suggested that the quantity of meibum secreted, and not quality, have a role in the onset of ARMGD, finally, they proposed ARMGD can occur without hyperkeratinization [17].

b. Decreased MG basal cell proliferation with aging

Studies have shown that with age, acinar basal cells of MGs present decreased proliferative potential [11,17,18,30,92]. Jester and colleagues have shown that from 1 year of age, mice start to display a decrease in the number of proliferating MG basal cells which occurs concomitantly with a decrease in acini size [18]. Ki67, the nuclear antigen present from G1 through mitosis stage of cell cycle but absent in resting cells, is widely used as a marker of the dividing cells. A labeling index calculated based on the number of Ki67 stained nuclei per 100 total nuclei in the basal cell layer of MGs decreased from 24.79 ± 7.77 in young (2 months old) mice to 11.28 ± 5.62 and 7.1 ± 10.04 in 1 year and 2 years old mice respectively [18]. In humans, a significant negative correlation was observed between the Ki67 labeling index and age, with a correlation coefficient of −0.591, (r 2 = 0.35; P < 0.001) [17]. Given that MGs produce meibum in a holocrine manner, a decrease in cell proliferation within the basal layer would lead to a reduced number of cells moving into the MG to initiate meibocyte differentiation and lipid accumulation, and, consequently, reduced meibum production. Also, with aging, the subcellular localization of PPARγ shifts from the cytoplasm to nucleus. Thus, taken together, these studies show that reduced proliferation and altered meibocyte differentiation contribute towards acinar atrophy, gland dropout, and decreased meibum secretion in ARMGD, in both mice and humans. Immune-fluorescent computed tomography (ICT) based 3-D reconstructed eyelids from aged mice (2 years old) also showed absence of proliferative ductules in atrophic glands using Ki67 proliferation marker. Importantly, the underlying cause of the reduced basal cell proliferation still remains unknown, however, strategies targeted at maintaining or increasing the MG basal cell proliferation has the potential to slow or even prevent the progression of ARMGD.

c. Loss of MG progenitor cells with aging

Early studies relied on label retaining techniques in order to identify MG progenitor cells, of note studies by Olami et al. (2001) [93] and Parfitt et al. (2016) [14]. The first study to seek to identify MG progenitor cells made use of pulse labelling cycling cells with 3H-thymidine followed by a 28-day chase period in rats [93]. A fundamental characteristic of stem cells/progenitor cells is that they are quiescent during homeostasis, and thus they retain genetic labels for extended periods. This study identified label retaining cells were located around the opening of the acini [93]. Over a decade later, using the K5/H2B-GFP mouse label retaining model, two groups of label retaining cells were identified within the region at the acini and ductile interface. Curiously, this study identified that after 58 days wash out period, only 1 or 2 label retaining cells remain in each acini, indicating it is a very small pool of progenitor cells that maintain the MG. Importantly, this group was able to establish that the MG progenitor cells express both K5 and K14 [94]. Similarly, in the skin, progenitor cells within the basal layer of the epidermis express Keratin 5/14 (K5/14), and are highly proliferative. These K5/14 cells are derived from stem cells that are located in the interfollicular epidermis or the hair follicle [95,96]. Interestingly, a follow up study using the Confetti transgenic mouse with the K14 promoter showed that each acini is likely replenished by a single progenitor cell [97]. Further characterization of these label retaining cell populations showed they also express Sox9 (although not exclusive to progenitor cells), and do not express Blimp1 or PPARγ. In fact, PPARγ is a well-known transcription factor that regulates lipidogenesis, and thus can be used as a marker of meibocyte differentiation [38]. More recently, leucine-rich repeats and immunoglobulin-like domains protein 1 (Lrig1) and early growth response protein 2 (EGR2 or Krox20) were proposed as putative markers for MG basal progenitor cells [98–100]. Although great progress has been made in the pursuit of potential putative markers for MG progenitor cells in recent years, further research is still needed in this area. A recent review was dedicated to provide an overview of the progress made over the past two decades in identifying MG stem cells [101]. The lack of MG progenitor cell markers has hindered studies related to understanding how the MG stem cell/progenitor cell pools are maintained over time. However, it is believed that a loss of MG stem cells/progenitor cells over time directly leads to the exhaustion of the proliferative basal cells and MG atrophy, however further research is still needed in this area.

6. Transcriptomic studies of mice with ARMGD

The cause and mechanism of ARMGD is still largely unknown. An attempt to understand the pathogenesis of ARMGD was made by Parfitt et al. (2016), using comparative transcriptomic analysis of eyelids of 3 month (young) and 2 year (aged) old C57BL/6 mice [102]. A total of 698 genes were identified as differentially expressed considering more than two-fold significant differences in either young or old samples. Regardless of sex, young and aged mice selectively expressed 65 and 201 genes, respectively. Combining differentially expressed genes with the selectively expressed ones in young and old tissues, 131 candidates were identified in the young mice and 567 candidate genes were identified in the aged mice. The significantly altered genes and pathways in young mice were associated with MG development and cellular defense, whereas in older mice, they were associated with differentiation, cell signaling and cell cycle regulation. Fibroblast growth factor (FGF) and Wnt functions were found to be significantly altered in the aging mouse. No significant differences in the expression of cytokeratins 1 & 10 and small proline rich protein (Sprr1a) were identified between young and old mice, further supporting the view that hyper-keratinization is not an underlying cause of ARMGD. Interestingly, Dickkopf like acrosomal protein 1 (DKKL1), was found to be significantly downregulated in aged mice. Immunohistochemical staining based on the association of PPARγ with DKKL1 suggested a potential role of DKKL1 in meibocyte differentiation and meibum synthesis which indeed reduces with age, however further investigation is required. A similar transcriptomic profiling of MGs from humans and other species would be invaluable for further understanding the pathogenesis of ARMGD.

7. Preventive and therapeutic strategies for ARMGD

As mentioned above, there is a lack of understanding on the underlying molecular mechanisms behind the atrophic involution of MGs with age. Therefore, no causal therapies for ARMGD exist to date. However, in recent years great progress has been made developing potential therapeutic interventions using animal models, which are now relying on effective clinical trials for translation into the clinic for treatment of ARMGD. Sasaki et al. (2022), have proposed that ARMGD and accompanying EDED can be ameliorated by enhancing local steroidogenesis [103]. Nicotinamide adenine dinucleotide (NAD+)-dependent circadian 3β-hydroxyl-steroid dehydrogenase (3β-HSD), the enzyme responsible for biosynthesis of steroid hormones [104], is expressed as type 1 isoform, namely HSD3B1 in human MGs [105] and as Hsd3b6 in mice MGs [106]. 3β-HSD activity shows circadian rhythmicity and declines with aging, and MGs of Hsd3b6^ΔMG mice, which lack any detectable 3β-HSD activity, have significantly reduced size. In the aged MGs, declined 3β-HSD activity was shown to be caused by limited NAD + availability, and exogenous supplementation with bio-precursors of NAD+ was tested for their therapeutic role in ARMGD, as previously reviewed in Yang et al., 2023 [101]. Topical administration of NAD + bio-precursors containing eye drops in 21-month-old mice four times a day for 90 days during daytime, activated endogenous 3β-HSD and resulted in significantly larger glands, increased BrdU-incorporated proliferative acinar cells and reduced atrophy and evaporative dry eye, suggesting that restoration of local intracrine activity could ameliorate MGD [103].

Periplocin, a natural extract from the dry root of Periploca sepium traditionally used in Chinese herbal medicine, has been shown to improve ARMGD in mice by upregulating the Na/K-ATPase via the SRC signaling pathway [107]. The acinar and ductal epithelium of MGs express Na/K-ATPase, but its expression declines by 83.7 % in aged mice (15 months old) compared to young mice (3 months old). Mice treated with 250 μg of periplocin (reconstituted in 10 % DMSO + 40 % PEG 300 + 5 % Tween-80 + 45 % saline) via subcutaneous injection into the upper eyelid twice weekly for 2 weeks, showed reversal of the age-associated reduction of Na+/K + ATPase, by 2.17-fold. Moreover, there was a 64.58 % increase in the number of Ki67-positive cells and 113.56 % increase in ΔNp63-positive cells in aged MGs, suggesting that periplocin may increase the renewal of MG epithelial cells in aged mice. Additionally, periplocin treatment significantly increased meibum production. Periplocin was shown to exert its effects by increasing the phosphorylation level of SRC in murine MGs, which otherwise declines dramatically with age. Overall, periplocin is proposed to be a potential therapeutic intervention for ARMGD, which promotes the proliferation and lipid production of MGs via the SRC signaling pathway [107].

Pigment epithelium-derived factor (PEDF), a 50-kDa monomeric glycoprotein, was first identified in the conditioned media of human fetal retinal pigment epithelial cells [108]. Since then, PEDF has been found to be expressed in many organs with a vital role in organogenesis and homeostasis. PEDF is abundantly expressed in the basal layer of the MG, however, with age, PEDF expression decreases significantly. A single subconjunctival injection of a 29 amino acid polypeptide based on PEDF (named 29-mer) into aged mice (15 months old) effectively stimulated the acinar cell proliferation in the remaining MGs, revealed by an increase in ΔNp63, Lrig1 and BrdU positive cells, and increased meibum production, culminating in improved the tear film stability in aged mice [109]. Therefore, PEDF is another promising therapeutic molecule for ARMGD [109,110].

Recent work by Qu et al. (2023), shows that IL-1β has a potential role in the pathogenesis of ARMGD. Specifically, there is an increase in IL-1β levels within the terminal ducts of MGs in aged rats (2 years old) when compared to young rats (2 months old). IL-1β was shown to induce MGD by inhibiting MG cell differentiation, which is necessary for holocrine secretion, to induce keratinization, inhibit basal cell proliferation, suppressing PPARγ expression, and, consequently, lipid accumulation, upregulate cytokeratin 1 and MMP9 expression, and, finally, promote apoptosis by activating the p38 MAPK signaling pathway [77]. Studies have shown that p38 MAPK inhibitors, i.e., SB203580, are able to reverse many of the effects of IL-1β, and therefore has the potential for treating ARMGD.

Finally, the Reneker group has been able to induce spontaneous MG regrowth using triple transgenic mice of K14rtTA;tetOCre;Fgfr2^fl/fl (referred as Fgfr2^CKO) and a reporter mouse line, K14-rtTA;tetO-Cre; Rosa^mTmG. In this study, doxycycline-induced deletion of the fibroblast growth factor receptor 2 gene (Fgfr2) resulted in acinar and ductal MG atrophy [111,112] in young adult Fgfr2^CKO mice (2 months old) within 14 days, with varying grades of severity after one (dox-1x group) or two (dox-2x group) intraperitoneal injections of doxycycline (80 μg/gm of body weight). Thirty days post-induction, MG regeneration was observed in both the groups. Recovered MGs had normal morphology, except they were smaller in size in the dox-1x group, however the regenerated MGs were heterogenous and structurally abnormal in the dox-2x group, indicating compromised recovery following severe ductal atrophy. The post-atrophy regenerated acini express normal differentiation markers, such as fatty acid synthase (FASN) and PPARγ, and present abundant meibum production, indicating functionally active meibocytes exist in the regenerated MGs. Importantly, acinar replenishment after MG atrophy is shown to be highly dependent on the remaining ductules and an inverse correlation exists between ductal atrophy and acinar regeneration. This study raises the possibility that MGs could be regenerated in patients with MGD, however an important question remaining is whether atrophic MGs associated with ARMGD could be induced to regenerate. Although the reversal of MG atrophy and increase in MG area has been reported following certain treatments [113–116], the possibility of regenerating MGs after MG drop-out has still to be explored.

8. Conclusion and future perspectives

ARMGD is an underdiagnosed and undertreated disease that afflicts a significant proportion of the population, especially the aging population. The pathophysiology of ARMGD remains poorly understood, therefore developing effective treatments to slow or even prevent ARMGD is challenging. Aging is a major risk factor for MGD, however, unfortunately, little is known about the underlying causes of the structural and functional changes that occur over time. Although obstruction of the MG collecting duct was traditionally believed to be the underlying cause of MGD, it is likely not the only cause. Moreover, obstruction of the MG excretory ducts and orifice is proposed to not be a major cause of ARMGD. Reduced basal cell proliferation, reduced number of MG progenitor cells, and increased inflammatory cell infiltration all precede MG atrophy; however, what drives these changes remains to be established. MGD is a major cause of DED, which significantly affects a patients quality of life, decreases productivity, and places them at higher risk for other ocular complications. Unfortunately, currently, solely palliative care is available for treating MGD and improving meibum production. Although different treatments for ARMGD, such as increasing local steroidogenesis, using p38 MAPK inhibitors, injecting periplocin or PEDF, have been proposed in recent years for treating ARMGD, these treatments have yet to be translated into the clinic. The eyelid and tarsal plate also undergo structural and biomechanical alterations caused by changes in the extracellular matrix (ECM) with age [117–119] thus, ECM based therapies also hold great potential for treating/preventing ARMGD. Taken together, there is still a pressing need for research geared towards understanding the underlying causes of MGD and ARMGD and how to prevent, or possibly even revert, MG atrophy.