ABSTRACT
ABSTRACT
Purpose/Aim: Dry eye disease (DED), common and suboptimally treated, is in need of novel animal models to understand its pathophysiology and assess the efficacy and other parameters of new pharmacological agents for its treatment. The more than 10 rabbit models of DED described to date have significant limitations including induction of mild disease, lack of consistency, and off-target effects when chemical agents are used for disease induction. Our aim was to develop a new model of chronic DED in rabbits that overcomes the limitations of existing models.
Materials and Methods: We performed a complete surgical resection of all orbital lacrimal glands (LGs; dacryoadenectomy) in normal adult New Zealand White rabbits. One week after removal of the nictitating membrane, we surgically removed the orbital superior LG, followed by removal of the palpebral superior LG, and finally removal of the inferior LG. Surgery was performed under anesthesia, required about 1 h/eye, and was well-tolerated.
Results: Dacryoadenectomy induced severe DED, evidenced by >90% reduction in the tear break up time test, 50% reduction in the Schirmer tear test, 10% increase in tear osmolarity, and a marked increase in the rose bengal staining score. DED was sustained and essentially unchanged for the eight weeks of observation. Sham-operated rabbits showed no such changes, with the exception of a non-significant and transient reduction in the tear break up time test, a response to ocular surgery.
Conclusions: This model of stable, chronic, predominantly aqueous-deficient DED recapitulates key clinical and histological features of human DED and is suitable for the study of ocular surface homeostasis, of the pathophysiology of DED, and of the efficacy of candidate drugs for DED treatment.
Introduction
Dry eye disease (DED) is one of the most commonly encountered problems in ophthalmology clinics, with a quarter of the patients reporting related symptoms.1 Multifactorial etiologies and varying clinical severity typify DED, which can result from pathology in any portion of the lacrimal functional unit.2 As their name implies, the lacrimal glands (LGs) are critical to tear production, and provide the majority of the aqueous layer, the middle and largest of the three layers of tears;3 the other two are the inner mucin coating and the lipid overlay.4 The pathophysiology of DED is conceptualized as deriving either from under-production or over-evaporation of tears. Sjögren’s syndrome, an extensively studied prototypical cause of DED, affects primarily the LGs and is a striking example of their importance in the pathogenesis of DED.
Given the enormous clinical importance of DED and the projected marked increase of its worldwide prevalence,4–7 the search for efficacious and safe therapeutic agents is of great import. As with most drug development efforts, informative animal models of DED are a crucial investigative tool. Notwithstanding the axiomatic statement that no animal model completely recapitulates a human disease, the plethora of DED animal models suggests that none are entirely satisfactory. Mouse, rat and rabbit animal models of DED are the most commonly used while dogs and primates are used infrequently.8–13
Murine models are attractive due to their lower cost, ability to create knockout models, and availability of reagents to study signaling and inflammatory molecules. However, other aspects of their physiology and anatomy still limit their use in drug development. For example, the much smaller size of their eyes, compared to human and rabbit eyes, makes drug biodistribution studies less reflective of the human, while the dissection of eye tissues is challenging. Rabbit models of DED, on the other hand, offer significant advantages over other species. The size of the globe, the well-defined eyelid anatomy, the level of expression of drug-inactivating carboxylesterases, and the LG histology of rabbits are much closer to those of the human than the eyes of rats and mice.8,14–16
At least 12 novel rabbit models of DED have been reported. The majority of them attempt to reduce tear production by either removing LGs or impeding their function. The most direct approaches include partial surgical resection of the LG (with or without concurrent removal of the nictitating membrane and Harder’s gland) or surgical closure of the LG excretory ducts with cautery.17,18 Impairing LG function has been done by the following means: irradiation of the LGs19; induction of dacryoadenitis by injecting the LGs with activated lymphocytes or the plant mitogen Concanavalin A targeting one or all orbital LGs;20–22 injection of botulinum toxin A to the palpebral portion of the superior LG;23 or LG denervation.24 Pharmacological agents such as topical atropine or benzalkonium chloride have also been used to induce primarily aqueous deficient DED in rabbits.25,26 Other methods include closing the Meibomian gland openings with cauterization,27 acute desiccation of the eye by manual prevention of blinking, and orchiectomy that depletes androgens that are required for tear production and for the normal structure and function of the corneal epithelium.28,29
These models recapitulate many features of aqueous deficient and/or evaporative DED. However, none of them completely abolish tear production from the orbital LGs. While irradiation of the LGs, administration of pharmacological agents and our dacryoadenitis model target all of the orbital LGs, the suppression of tear production, often inconsistent in its degree, remains incomplete. Partially impaired LGs may still produce significant tear volumes with at times compensatory tear overproduction by residual LGs.22 Surgical models that involve partial resection of the orbital LG tissues,15,17,18,30,31 as well as surgical denervation of the LG, are limited by significant residual tear production.24 The origin of the residual tears in these models remains uncertain; possible sources include the unresected LGs, the accessory lacrimal glands, and residual orbital LG tissues, each alone or in combination.
Prompted by the need for a model that completely and reliably eliminates all tear production from the orbital LGs, we developed a practical method for their complete surgical removal in the rabbit. Herein, we describe this new technique and highlight its distinct advantages over previously described models. This model could be useful in studies of the contribution of the orbital LGs to basal tear production or of the homeostasis and pathophysiology of the ocular surface in aqueous-deficient DED. In addition, this model, chronic and stable, might be used in the evaluation of DED therapeutics.
Animals and methods
Study animals
Male New Zealand White rabbits (NZW), each weighing 2–3 kg, were housed singly in a room with a strict 12-h on/off light cycle, temperature of 65 ± 5° F, and humidity of 45 ± 5%. During experiments, animals had free access to standard chow and water without nutritional enrichments that might affect tear production or stability. Rabbits were acclimated to this environment for ≥2 weeks prior to any study. All animal studies were in accord with the Statement for the Use of Animals of the Association for Research in Vision and Ophthalmology and approved by our relevant institutional committees.
Determination of DED
We assayed several parameters to determine DED in rabbits; the first three below were recently described in detail.22 For all measures, each animal was placed in a restraining bag and sedated mildly with acepromazine (1–2 mg/kg subcutaneously). Measures were taken 5–10 min after sedation was given; at that point, all endpoints of sedation (e.g., relaxed body tone, lowered ear position and decreased head movements) had been reached. All measures were done in the same order as described below and at the same time of day to minimize the chance of circadian variations effecting results. Measures were taken at baseline and then at postoperative weeks 1, 2, 3, 4, 6 and 8 in the dacryoadenectomy group. Animals undergoing sham surgery were measured at baseline, and then at postoperative weeks 2, 3, 4 and 6 for all parameters except rose bengal staining, which was scored at baseline and postoperative weeks 1, 2, 3, and 4.
Tear osmolarity
After sedation and gentle retraction of the lower lid, tears are sampled with the TearLab Osmometer (TearLab Corp., San Diego, CA) and osmolarity is measured following the manufacturer’s instructions.
Tear break up time (TBUT)
Following topical anesthesia and the placement of a wire lid speculum, a 50 µl drop of 0.2% fluorescein is evenly distributed over the eye, and the pre-corneal tear film is observed under blue light with surgical loupes (Designs for Vision, Ronkonkoma, NY). The time taken to develop black dots, lines or obvious disruption of the fluorescein film is measured up to 1 min. If break up is not seen by 1 min, observation is halted and TBUT is recorded as 60 s (even if actually longer).
Schirmer tear test (STT)
After a 50-µl drop of 1% preservative-free lidocaine is instilled on the eye, a Weck-Cel® surgical sponge (Beaver-Visitec International, Waltham MA) is placed into the lower fornix to remove all fluid. After 20 s, the surgical sponge is removed and Color Bar Schirmer strips (EagleVision, Memphis TN) are placed in the mid portion of the lower lid. Tear production is recorded as the length of moistened strip at 5 min. The final score is the average of three separate measurements.
Rose bengal staining
After instillation of 50 µl of 1% preservative-free lidocaine using a micro-pipette, 25 µl of rose bengal 1% is instilled on the ocular surface, the eyelid is manually blinked once to evenly distribute it, and a timer is started. At 4 min, a wire lid speculum is placed and photographs of the superior conjunctival and corneal surface are taken with the same camera and illumination.32,33 We used a Canon 5D camera with the following settings: Aperture priority mode (aperture 8), ISO 6000, Canon 100 mm Macro lens attached with two 12.5 mm extension tubes, manual focus mode, lens at maximum magnification, and illumination from Canon Ring MR-14EXII flash set to automatic with ETTL mode. Pictures are taken with a ring flash’s focus lamp activated to aid focus on the cornea. Photos for both eyes are completed within 1 min.
Images were graded using a modified NEI method.34 Rather than grading six separate conjunctival zones, we only scored the superior conjunctiva of each eye, the only area easily photographed without manipulating the globe, which could artifactually change the ocular surface. Photos of the superior conjunctiva were scored by two ophthalmologists blinded to their identity; their scoring had a correlation coefficient of 0.97.
The complete dacryoadenectomy method
Rabbits were randomly assigned to the dacryoadenectomy group (n = 8) or to the sham-operated group (n = 2). In all animals, all surgical procedures described below were performed bilaterally. All surgeries were done by a team of at least two investigators. All surgery was performed by an ophthalmologist (RH) with the assistance of an experienced laboratory animal investigator (LH).
The anatomy of the orbital LGs of the rabbit
The anatomy of the orbital LG of the rabbit has been classically described by Davis.35 Figure 1a depicts the location of the two orbital LGs of the rabbit. The smaller superior LG (SLG), is composed of the orbital (OSLG) and the palbebral (PSLG) portions. The OSLG resides medially, deep within the superior orbit, and cannot be directly visualized or safely removed through cutaneous incisions around the temporal or inferior orbit. The PSLG, located in the temporal portion of the upper lid, appears as a small bulge when the upper lid is everted. The significantly larger inferior LG (ILG) extends along the entire inferior orbit from its anterior to posterior limit, with its posterior half lying underneath (medial to) the zygomatic bone.
Published online:
15 April 2019Figure 1. The rabbit orbital lacrimal glands.
a. The superior and inferior lacrimal glands (LGs) are depicted. Part of the inferior LG lies underneath the zygomatic bone. Upper right: orientation coordinates. b: Radiographic identification of the position of the LGs in axial and sagittal views of the head of a rabbit. The five radiopaque localization markers placed on the skin are clearly visible. The radiographic contrast dye provides the outline of the two LGS, shown also in the diagrams, and confirms the very medial location of the orbital superior LG in the orbit.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0001_oc.jpg"})
Figure 1. The rabbit orbital lacrimal glands.
a. The superior and inferior lacrimal glands (LGs) are depicted. Part of the inferior LG lies underneath the zygomatic bone. Upper right: orientation coordinates. b: Radiographic identification of the position of the LGs in axial and sagittal views of the head of a rabbit. The five radiopaque localization markers placed on the skin are clearly visible. The radiographic contrast dye provides the outline of the two LGS, shown also in the diagrams, and confirms the very medial location of the orbital superior LG in the orbit.
Radiologic studies confirmed the close proximity to the globe of the SLG and ILG, in keeping with their physiological role.36 Five radiopaque balls, placed on the skin overlying the right SLG and ILG, served as position markers. The LGs of the right eye were surgically exposed and injected under direct visualization with Omnipaque 300 radiographic contrast dye (GE Healthcare, Wauwatosa, WI), and radiographs were obtained (Figure 1(b)). These radiographs confirm the very medial location of the OSLG in the orbit, which cannot be removed through previously described attempts using temporal or inferior periorbital transcutaneous incisions. We were, however, able to completely and safely remove the OSLG and all other orbital LGs by exploiting the rabbit’s cranial anatomy as described below.
Removal of the nictitating membrane
After sedation with acepromazine 1mg/kg, a wire lid speculum is placed between the lids. The nictitating membrane (NM) is infiltrated with 2% lidocaine with 1:100,000 epinephrine and gently pulled away from the globe and cut off with scissors close to its base. Care is taken not to cut off the NM too closely to its base, to avoid Harder’s gland prolapsing into the subconjunctival space. Cautery is generally not necessary, and its occasional use is minimal. Polysporin ophthalmic ointment is applied to the eye once. Further ocular procedures, or measurements are not performed for at least one week until clinical exam shows that the ocular surface has healed completely. Harder’s gland, adjacent to large venous sinuses in the orbit, is not removed to avoid the risk for massive bleeding.
Removal of the LGs
Figure 2 shows the surgical steps. All fur is first removed using shears and Nair® to make more obvious the contours of the LGs. Surgery is performed under general anesthesia with Xylazine/Ketamine induction and maintenance with isoflurane, using an R3 V-gel (Docsinnovent/Jorgensen Laboratories, Loveland CO) to maintain the airway. Respiration, heart rate, respiratory tidal volume, and arterial oxygen saturation are monitored throughout the procedure. Animals are kept on a heating pad to prevent hypothermia. A reverse Trendelenburg position is maintained to lessen the chance for excessive bleeding. Surgical incisions are identified and marked with a pen. The anterior portion of the ILG appears as a protuberance in the skin under the anterior portion of the eye. The OSLG location is identified by gently applying medial pressure to the globe and looking for the area on the skull which bulges, typically in line with the posterior (lateral) canthus and about 1 cm medial to the bony edge of the superior orbital rim. The periorbital area, scalp, and globe are prepped with a 10% povidone-iodine solution diluted to half strength with normal saline and then draped to maintain a sterile field.
Published online:
15 April 2019Figure 2. The steps of the surgical resection of the lacrimal glands.
a: Subcutaneous lidocaine injection for anesthesia. b: Removal of OSLG through a transcranial approach on top of the skull. c-e: Transconjunctival approach to remove PSLG after upper eyelid is everted. f-k: Removal of the ILG through transcutaneous approach below the lower lid; the conjunctival space is not violated with the removal of ILG. h: Anterior reflection of the ILG showing the zygomatic bone. i: Eye following surgical closure.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0002_oc.jpg"})
Figure 2. The steps of the surgical resection of the lacrimal glands.
a: Subcutaneous lidocaine injection for anesthesia. b: Removal of OSLG through a transcranial approach on top of the skull. c-e: Transconjunctival approach to remove PSLG after upper eyelid is everted. f-k: Removal of the ILG through transcutaneous approach below the lower lid; the conjunctival space is not violated with the removal of ILG. h: Anterior reflection of the ILG showing the zygomatic bone. i: Eye following surgical closure.
The OSLG should be removed first. If it follows the removal of the ILG, its excision is considerably more difficult because it recesses deeper into the orbit and cannot be prolapsed as easily through the posterior incisure (Figure 2). An incision is made through the skin and subcutaneous tissues directly over the posterior incisure using a Colorado needle (Kalamazoo, MI 49002). Alternating medial pressure on and off the globe will cause the OSLG to prolapse and recess into the posterior incisure creating a visible change in the tissue contours that aids finding the optimal dissection location often just medial or deep to preauricular muscle fibers. Incising the remaining connective tissues over this bulge exposes the posterior incisure and the OSLG tissue, which is typically pale pink or tan in color. Using blunt, non-toothed forceps, the OSLG is harvested using traction to slowly tease the gland out from its deeper position within the orbit. This usually requires cutting small connective tissue bands as it is removed. As the OSLG is pulled out from the posterior incisure, its contour tapers as it transitions into the main excretory duct. After as much as possible of the gland and excretory duct are withdrawn, it is truncated as far inferiorly as possible using generous amounts of cautery. Cautery not only provides hemostasis but also serves to provide a visible darkening of the surrounding tissues that can often be seen when removing the superior portion of the PSLG as described below.
To remove the PSLG, the upper eyelid is everted. The termination of the PSLG is easily seen here as a small bulge in the tissue (Figure 3(a)). Gentle pressure on the lid creates blanching, sometimes making it possible to visualize the main excretory duct, which often appears as a more whitish band of tissue approximately 2 mm wide running through the lid. The gland tissue is grasped with forceps along the bulge and removed using sharp dissection and gentle traction to maintain a surgical plane. It is difficult to visualize a clear differentiation of the excretory duct and PSLG tissue, but the tissue plane is remarkably easy to maintain. As traction is applied, the dissection plane will advance posteriorly and superiorly to terminate in the superior temporal portion of the lid, where it is common to see cautery marks made when removing the OSLG described above (Figure 3(b)). Mild bleeding is self-limited and usually stopped with gentle pressure using a cotton swab. No sutures are needed for this portion.
Published online:
15 April 2019Figure 3. The palpebral portion of the superior lacrimal gland during surgery.
a: During surgery, the palpebral portion of the superior lacrimal gland (PSLG), demarcated by the arrows, can be seen as a bulbous protrusion in the most posterior portion of the upper lid after it has been everted. b: This photograph was taken during removal of the right PSLG. The freely dissected portion of the PSLG is shown between the blue arrows. This tissue plane terminates adjacent to the superior orbital rim. The yellow arrow points to cauterized tissue remnants of the orbital SLG already removed through the transcranial approach.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0003_oc.jpg"})
Figure 3. The palpebral portion of the superior lacrimal gland during surgery.
a: During surgery, the palpebral portion of the superior lacrimal gland (PSLG), demarcated by the arrows, can be seen as a bulbous protrusion in the most posterior portion of the upper lid after it has been everted. b: This photograph was taken during removal of the right PSLG. The freely dissected portion of the PSLG is shown between the blue arrows. This tissue plane terminates adjacent to the superior orbital rim. The yellow arrow points to cauterized tissue remnants of the orbital SLG already removed through the transcranial approach.
To remove the ILG, a curvilinear incision is made through the skin and the subcutaneous tissues surrounding the posterior and inferior orbit similar to (but more extensive than in) previously described methods.15,17,31,37 The posterior (lateral) canthus tendon is also exposed, facilitating the identification and removal of the posterior portion of the ILG, which lies partially under this structure in some animals. Care must be used in this region, as a branch of the external carotid is in close proximity to or in direct contact with, the terminal portion of the ILG. Attention should be given to ensure complete removal of the main excretory duct of the ILG as it passes through the connective tissues of the lower lid to end in the lower fornix. (This approach does not disrupt the conjunctival tissues of the lower fornix.) After the removal of all orbital gland tissues, the deeper tissues, including all muscle and connective tissues, are re-approximated with 4–0 Mersilene suture. The periorbital incisions are closed with a running 6–0 Vicryl suture that incorporates the skin and orbicularis muscle tissue. Large venous sinuses lie more medial to the inner portions of the ILG and great care must be taken to avoid inadvertent damage to this structure, which might result in massive bleeding.
Sham procedure
The sham procedure is performed using the identical transcutaneous incisions. The anterior portion of the ILG and SLG is exposed and their connective tissue capsules are opened but the glands are not removed or dislodged from their natural position. The connective tissue capsule of the posterior portion of the ILG is not opened because doing so often results in spontaneous protrusion of the gland, which alone might impact tear production. Skin and deep tissue closure is done as described above. Similarly, the PSLG is not removed during surgery. Instead, a small incision is made more anteriorly in the conjunctival tissues of the upper lid with a scissor.
Postoperative care
Following surgery, ophthalmic ointment is placed on the eye and over the skin incisions. Normal saline (20 ml) is given subcutaneously, anesthesia is reversed and rabbits are monitored following standard protocols. Using a careful surgical technique, these procedures can be done using only mild bipolar cautery. Although not needed, we keep hemostatic aides available. Animals requiring hemostatic aides would have been removed from the study because these aids can elicit an inflammatory response38 potentially impacting the DED inflammatory cycle.
Statistical analysis
Each outcome was analyzed across time using linear mixed models with time as a within-subjects factor and random subject effects. Post-hoc tests were conducted to determine the nature of significant effects. The best-fitting model was determined using information criteria. These models allowed for modeling the correlation between multiple measurements taken within the same animal. SAS, version 9.4 (Cary, NC) was used to perform all analyses. P values less than 0.05 were considered significant.
Results
Outcomes of the method
The complete surgical removal of all orbital LGs as described here required only a moderate degree of skill and the learning curve was not particularly steep, with ophthalmic surgeons acquiring proficiency with about five iterations. The completeness of the removal of the LGs was readily assured at surgery as all margins for both SLGs and the ILG were visualized. Total surgical time was about 2.2 h; this excludes the removal of the nictitating membrane done separately and requiring less than 10 min. Our method proved extremely safe without fatalities or intraoperative complications. In particular, no rabbit had significant bleeding and no hemostatic aid other than modest cautery was ever employed.
Surgical dissections produced LG specimens generally consistent with previous anatomical descriptions. The only exception concerned the size of the OSLG, located in the superotemporal portion of the orbit, reported to be round, flat and approximately 5 mm in size.35 In our study, the OSLG specimens were frequently larger (about 1 cm); the resected tissue was confirmed histologically to be LG (Figure 4). Surgical manipulation and traction applied to the tissue during its removal through the posterior incisure may account for the discrepancy to some extent. The PSLG specimens, whose surgical resection has not been previously used to induce DED, and the ILG specimens were consistent with descriptions by Davis and others.8,35
Published online:
15 April 2019Figure 4. The surgically resected rabbit lacrimal glands.
Histological tissue sections taken from the LGs stained with hematoxylin and eosin. The SLG (a) and ILG (b) both demonstrate typical acini in a tubuloalveolar architecture. c: The most terminal portion of the PSLG located in the posterior part of the superior lid. Red arrow: A large excretory duct lined by cuboidal cells as well as acini in a glandular pattern is seen. Green arrow: A blood vessel. There is also sebaceous tissue nearby from a Meibomian gland.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0004_oc.jpg"})
Figure 4. The surgically resected rabbit lacrimal glands.
Histological tissue sections taken from the LGs stained with hematoxylin and eosin. The SLG (a) and ILG (b) both demonstrate typical acini in a tubuloalveolar architecture. c: The most terminal portion of the PSLG located in the posterior part of the superior lid. Red arrow: A large excretory duct lined by cuboidal cells as well as acini in a glandular pattern is seen. Green arrow: A blood vessel. There is also sebaceous tissue nearby from a Meibomian gland.
Table 1 summarizes the weight of the resected LGs. As expected, the ILG is on average more than sixfold larger than the SLG. Of the two portions of the SLG, the OSLG is sevenfold larger than the PSLG. The SLG contributes only 14.4% of the total weight of the LG tissues, the rest being represented by the ILG. We observed considerable variation in the mass of both the SLG and ILG glands. The range in mass of the ILG was 2.3-fold (564 mg to 1325 mg) and that of the SLG 2.7 fold (104 mg to 276 mg). These findings are also consistent with the indirect assessment of size of the ILG by ultrasonography that we recently reported22; that study emphasized the variability in the anatomical size of the ILG with respect to the zygomatic bone.
Table 1. The weight of the excised rabbit lacrimal glands.
Induction of dry eye
Our approach successfully accomplished its main objective, the induction of dry eye. This was confirmed by characteristic changes in clinical and laboratory parameters of DED; their values on postoperative week 2 are summarized in Table 2. Impression cytology was performed on all eyes at baseline and during the follow-up period. All eyes undergoing complete dacryoadenectomy exhibited a marked reduction in goblet cell numbers and epithelial changes consistent with dry eye as shown in Figure 5 (details to be published elsewhere).
Table 2. Dry eye parameters on postoperative week 2.
Published online:
15 April 2019Figure 5. Conjunctival impression cytology demonstrating loss of goblet cells following complete dacryoadenectomy.
Representative impression cytology from the superior conjunctiva that has been stained with PAS and hematoxylin (magnification 100x). Panel A shows numerous, plump goblet cells at baseline (after acclimatization). Panel B taken 1 month following complete dacryoadenectomy shows complete loss of goblet cells from the same area sampled in panel A. The underlying epithelial cells from the conjunctival surface are still present. All animals undergoing surgical dacryoadenectomy showed similar, or more severe, changes consistent with DED.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0005_oc.jpg"})
Figure 5. Conjunctival impression cytology demonstrating loss of goblet cells following complete dacryoadenectomy.
Representative impression cytology from the superior conjunctiva that has been stained with PAS and hematoxylin (magnification 100x). Panel A shows numerous, plump goblet cells at baseline (after acclimatization). Panel B taken 1 month following complete dacryoadenectomy shows complete loss of goblet cells from the same area sampled in panel A. The underlying epithelial cells from the conjunctival surface are still present. All animals undergoing surgical dacryoadenectomy showed similar, or more severe, changes consistent with DED.
Reduction of TBUT
At postoperative week 1, complete surgical dacryoadenectomy reduced TBUT by more than 90%, compared to preoperative levels (Figure 6(a)). Mean TBUT remained suppressed by more than 75% of preoperative levels (p < 0.0001 for all time points) during the 8 weeks of observation. A trend for a very slow gradual improvement might have been present, although the changes are not statistically significant. Rabbits undergoing the sham procedure had a 5–47% decrease in TBUT, which gradually returned to normal by the 6th postoperative week. However, the decreases in TBUT following sham surgery were much less than in the dacryoadenectomy animals, never reaching statistical significance.
Published online:
15 April 2019Figure 6. The response of DED parameters to dacryoadenectomy.
Changes in TBUT (a), Schirmer tear test (STT) (b), osmolarity (c), and rose bengal staining of the ocular surface (d) in dacryoadenectomized and sham-operated NZW rabbits. Values: mean ±SEM; n = 16 eyes/group (eight rabbits), except for weeks 7 and 8 (n = 5 rabbits).
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0006_oc.jpg"})
Figure 6. The response of DED parameters to dacryoadenectomy.
Changes in TBUT (a), Schirmer tear test (STT) (b), osmolarity (c), and rose bengal staining of the ocular surface (d) in dacryoadenectomized and sham-operated NZW rabbits. Values: mean ±SEM; n = 16 eyes/group (eight rabbits), except for weeks 7 and 8 (n = 5 rabbits).
Suppression of STT
As shown in Figure 6B, complete dacryoadenectomy resulted in a rapid and significant decrease in STT that remained stable for the 8 weeks of observation. STT showed no trend for recovery during the entire follow up period. Mean STT remained decreased approximately 50% across all postoperative time points, with a peak reduction in mean STT of 58% (18.3 to 7.6 mm) at post-op week 5. As expected, sham animals showed no significant decrease in STT at any of the postoperative visits.
Increased tear osmolarity
The baseline tear osmolarity was 291.3 ± 37 mOsm/L (mean±SEM for this and subsequent values). Tear osmolarity increased approximately 10% in response to dacryoadenectomy, starting on the first postoperative week (314.8 ± 2.8 mOsm/L) and remained elevated throughout the entire postoperative period (Figure 6(c)). The highest tear osmolarity value was reached on postoperative week 8 (325.7 ± 3.5 mOsm/L) representing an increase of 12% from baseline. Sham-operated animals did not show a statistically significant increase in tear osmolarity following intervention, measuring 299 mOsm/L at baseline and about 303 mOsm/L across all its postoperative measurements.
Increased rose bengal staining
At baseline, there was no evidence of any rose bengal staining (modified NEI score = 0) for all animals. Complete dacryoadenectomy resulted in a statistically significant increase in rose bengal staining with an average score of 5.4 across all time points measured after surgery (p < 0.001), as shown in Figure 6d and Figure 7. The mean modified NEI score remained elevated for the duration of the study, without signs of recovery. The highest scoring was seen at week 6 postoperatively with a mean modified NEI score of 6.6. Sham-operated animals also showed a much smaller increase in rose bengal staining following the procedure, peaking at postoperative week 2 (2.25 ± 0.47).
Published online:
15 April 2019Figure 7. Rose bengal staining of corneal surface following dacryoadenectomy.
a: Normal eye at baseline, without any staining defects. b: One week following removal of all LG tissues, the corneal surface is markedly changed from the induced dry eye showing irregularity of light reflex, increased punctate and linear staining of the corneal surface, and small mucous accumulation. c: Two weeks following surgery, there are similar changes, although the light reflex and overall staining have improved slightly. All images were taken 4 min after the application of rose bengal.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/icey20/2019/icey20.v044.i08/02713683.2019.1594933/20190802/images/medium/icey_a_1594933_f0007_oc.jpg"})
Figure 7. Rose bengal staining of corneal surface following dacryoadenectomy.
a: Normal eye at baseline, without any staining defects. b: One week following removal of all LG tissues, the corneal surface is markedly changed from the induced dry eye showing irregularity of light reflex, increased punctate and linear staining of the corneal surface, and small mucous accumulation. c: Two weeks following surgery, there are similar changes, although the light reflex and overall staining have improved slightly. All images were taken 4 min after the application of rose bengal.
Discussion
Our study describes for the first time a complete dacryoadenectomy method in rabbits that uniformly produces DED. In this model, DED, documented using criteria assessing several aspects of its pathophysiology, results from aqueous tear deficiency leading to secondary changes on the ocular surface. DED is pronounced and sustained, establishing this model as one of chronic DED.
Previous attempts to establish DED through incomplete dacryoadenectomy in rabbits have not removed the superior OLG.15,17,18,30,31,39 After numerous dissections of necropsy samples, we found it impossible to remove what we describe as the superior OLG through the periorbital incisions described in these reports. As we have, however, recently demonstrated, the SLG makes an important contribution to tear production.22 Thus the development of a successful dacryoadenectomy model necessitates the removal of this gland in its entirety. Our method provides a safe, simple and efficient (even though counterintuitive) approach to the removal of the SLG. As practiced, the removal of the SLG is accompanied by minimal blood loss and avoids bone resection with its attendant potential for damage to the eye. In addition, the transcranial approach to the SLG minimizes damage to the ocular surface tissues, an important point for DED, an ocular surface disease.
The presence of functional lacrimal tissue along the main excretory duct of the rabbit lacrimal system has been reported.35 The direct visualization of the LGs afforded by our method makes easy the removal of these important, even though small in size, portions of the lacrimal gland system. We always take care to remove them, in order to preclude the compensatory production of tears beyond their baseline capacity.
The ILG presented a challenge. This gland is partially covered by the zygomatic bone, and its size can vary by as much as fourfold.22 In addition, the posterior portion of the gland lies in close proximity to branches of the external carotid artery, while the anterior portion lies immediately adjacent to large venous sinuses. Damage to either of these vascular structures could result in massive bleeding. Our approach, allowing for complete visualization of the gland as well as these vascular structures ensures complete removal of the ILG and essentially eliminates the risk of vascular trauma.
The resection of the two LGs was complete with our method. This conclusion is supported by the observation that dacryoadenectomy produced a state of sustained DED. The LGs likely have the capacity to grow after partial removal or can, to some degree, compensate for their partial destruction. This was shown with the Concanavalin A-based model of DED, where a partially inactivated ILG overproduced tears.22 With this technique, the accessory LGs are the only remaining source of tears.
The learning curve for our method is fairly short, despite the procedure appearing arduous; ophthalmic surgeons acquired proficiency with about five iterations. About 1 h allows for the comfortable completion of the dacryoadenectomy of one eye, with no undue stress being placed on either the animal or the surgical team. Therefore, we consider this to be an easily applicable method. Nevertheless, it merits mentioning that prior experience in orbital surgery will greatly accelerate mastering these procedures.
The induction of DED by complete dacryoadenectomy was firmly established. Four parameters of DED, TBUT, tear osmolarity, the Schirmer tear test and rose bengal staining, gave congruent results consistent with the induction of severe DED.
The TBUT reduction, the most dramatic, persisted throughout the study period. In contrast, sham-operated animals had a statistically non-significant reduction in TUBT, with full recovery by the 6th post-operative week. As TBUT can be affected by alterations of the ocular surface, this change in TBUT for sham-operated eyes is not surprising, nor is its variability. In fact, similar reductions in TBUT have been documented in humans due to the surgical trauma,40 following LASIK/PRK41, and lamellar keratoplasty.42
The tear osmolarity of the dacryoadenectomized rabbits had an immediate (1 week) and dramatic average increase of 29 mOsm/L at all post-operative measurements and persisted unabated for at least 8 weeks. Both the baseline value and the increased postoperative levels of tear osmolarity are consistent with previous reports and reflect the values seen in human DED.18,43–45 Sham-operated animals displayed practically unchanged normal tear osmolarity values.
Similar to the previous two parameters, STT showed a rapid response to dacryoadenectomy with a marked (on average 50%) and stable reduction, apparent at the first measurement on postoperative day 7. The sham-operated animals showed no such change, remaining overall practically unaffected. Partial dacryoadenectomy (removal of the ILG alone) is reported to reduce STT by only 32% on average,17,30 likely reflecting the contribution of the SLG left behind.
Complete suppression of STT post dacryoadenectomy did not occur in our model. Failure to completely ablate all tear production with lacrimal gland removal is not surprising and has been documented in multiple other species including mice,46 rat,47 rabbits,48 squirrel monkey,13 and even humans.49 Although the complete removal of all LG tissue in most of these reports is unlikely, the definitive source of this residual tear production remains elusive. In our model, careful anatomic dissection at necropsy showed tear production was not due to regrowth of any LG tissues. Other possible sources for the residual tear fluid include the largely untouched accessory LGs, fluid from other conjunctival sources and plasma leakage from conjunctival vessels.50–52 Previous works have provided evidence that conjunctival and/or corneal tissues have the ability to secrete or transport fluid into the tear film.48,51,53,54 Changes in aquaporin expression and/or function may underlie some of the residual tear production.15 Harder’s gland is an unlikely source as it produces a small amount of a predominantly lipid-based emulsion.35,55 Because our simple and safe surgical method completely ablates all of the orbital LG tissues, our model provides an excellent means to study the in vivo compensatory mechanisms of the conjunctival and corneal tissues in a setting of aqueous deficient DED.
In dacryoadenectomized rabbits, rose bengal was markedly increased, reaching an essentially stable high value on its first measurement at postoperative day 7. The sham-operated animals, responding to the incision on the upper palpebral conjunctiva, displayed a very modest increase in the NEI score, which started quickly returning to baseline as the conjunctiva was healing.
Rose bengal, a derivative of fluorescein introduced by Sjögren himself, has selective staining patterns in dry eye or Sjögren’s syndrome.56,57 Rose bengal stains areas with breaks in tear-film integrity or altered tear-film components such as mucin, which normally act as a barrier against rose bengal. Thus, our findings indicate a significant change in dacryoadenectomized rabbits in some of these protective components that can no longer effectively shield ocular epithelial cells from the dye molecule. We have indeed observed changes in corneal mucin levels in rabbits with dry eye (to be published elsewhere).
That no animals developed severe ocular surface changes such as epithelial breakdown or vascularization, suggests as yet unidentified protective mechanisms, presumably operating to maintain corneal clarity. This is a critical aspect of this model, especially for its use for drug efficacy studies. Our complete dacryoadenectomy model presents an excellent system for mechanistic studies of the ocular surface in the context of either acute or prolonged aqueous deficient states.
Our method differs significantly from previous attempts at induction of DED by removing the contribution of the orbital LGs to tear production in the rabbit. None of the reported rabbit models eliminated tear production by all of the orbital LG tissues. The five known surgical models describe the removal of only the ILG.15,17,18,30,31 The partial efficacy of these methods is evidenced by the modest reduction in STT, which averaged 32% and ranged between 8.5% and 40%. The reasons for the difference between the present method and the previous ones are the unimpeded tear production by the intact SLG and the probably at times incomplete removal of the ILG in the latter. Similarly, parasympathetic denervation of the lacrimal glands is not able to fully suppress tear production,24 which is stimulated by neurocrine, endocrine, autocrine or paracrine factors.36 Finally, pharmacological suppression of tear production has been incomplete while off-target effects of the agents being used might complicate data analysis.25,26 Thus, our method offers distinct advantages over existing methods. It is also very likely that resection of only one of the glands, full or partial, following our thorough approach could generate DED of varying severity, mimicking the spectrum of its clinical manifestations.
In conclusion, our data demonstrate that complete dacryoadenectomy in the rabbit produces an animal model of stable, chronic, predominantly aqueous deficient DED that recapitulates key clinical and histological changes of human DED. This model is suitable for the study of ocular surface homeostasis, of the pathophysiology of DED, and of the efficacy and safety of candidate drugs for its treatment.
| Palpebral Superior LG | Orbital Superior LG | Inferior LG | All LGs | |
|---|---|---|---|---|
| weight, g; n = 16 eyes | ||||
| Mean ± SEM | 21.4 ± 1.5 | 151.7 ± 52.3 | 1,027.0 ± 53.5 | 1,200.1 ± 58.3 |
| Range | 10–32 | 94–244 | 564 −1325 | 680–1536 |
| Median | 21.5 | 143.5 | 1,043.0 | 1,238.5 |
| Sham operation | Dacryoadenectomy | |||
|---|---|---|---|---|
| mean ± SEM; n = 16 eyes | ||||
| Baseline | Week 2 | Baseline | Week 2 | |
| Tear break-up time, sec | 47.5 ± 12.5 | 31.8 ± 16.3 p = 0.47 | 60.0 ± 0.0 | 4.5 ± 1.2 p < 0.0001 |
| Tear osmolarity, mOsm/L | 299.9 ± 3.2 | 296.3 ± 4.1 p = 0.51 | 291.2 ± 3.7 | 315.3 ± 5.5 p = 0.001 |
| Schirmer tear test, mm | 17.2 ± 3.4 | 16.4 ± 2.3 p = 0.37 | 18.3 ± 1.3 | 10.5 ± 1.6 p = 0.0006 |
| Rose bengal, modified NEI score | 0.0 ± 0.0 | 2.25 ± 0.5 p = 0.003 | 0.0 ± 0.0 | 4.28 ± 0.6 p < 0.0001 |
Operated vs. baseline: Dacryoadenectomy: TBUT, p < 0.0001; tear osmolarity, p < 0.001; Schirmer tear test, p < 0.0006; and rose bengal. Sham operation: All not significant.
Acknowledgments
The authors wish to thank Michele McTernan for editorial support for preparation of the manuscript and Dr. Nengtai Ouyang for helpful comments.
Disclosure Statement
The authors declare no conflict of interest except for BR who has an equity position in Medicon Pharmaceuticals, Inc. and Apis Therapeutics, LLC, and LH, who is an employee of Medicon Pharmaceuticals, Inc. and has an equity position in Apis Therapeutics, LLC.