Introduction
In recent decades, deep anterior lamellar keratoplasty (DALK) has become increasingly popular for the treatment of corneal pathologies with a healthy endothelial cell layer. One of the major
advantages over penetrating keratoplasty (PKP) is the avoidance of an endothelial immune response, which is particularly beneficial for young patients with advanced keratoconus [1 ], [2 ]. The surgical procedure of DALK includes a variety of different approaches that affect the postoperative visual outcome [3 ], [4 ]. Most notably, removal of the recipientʼs corneal stroma down to Descemetʼs membrane (DM), so-called descemetic DALK (dDALK), has been
shown to provide excellent optical quality of the interface, resulting in outcomes comparable to PKP [5 ].
Nevertheless, little is known about the cellular responses and remodeling processes of the interface area after DALK, which might affect the long-term morphological and functional outcomes.
Especially in case of remaining recipient corneal stroma (predescemetic DALK, pDALK), the interaction between both stromal sides may be accompanied with a higher incidence of interface
haziness [6 ].
In this case report, we present two patients who underwent PKP after pDALK. The corneal buttons were further examined with light and transmission electron microscopy (TEM) to visualize the
interface area and highlight the ultrastructural corneal changes after DALK.
Case Report: Patient History and Clinical Presentation
Case 1
A 58-year-old male patient initially presented to our department due to distinct visual deterioration in the right eye. He underwent DALK previously because of corneal stromal opacification
of unknown origin. According to the patientʼs history, visual acuity was satisfying for years after surgery. One year after, cataract surgery was carried out without any complications. Due
to high regular astigmatism, limbal relaxing incisions were performed 4 years after, resulting in rapid visual deterioration with monocular diplopia. At the initial examination at our
department, the right eye showed significant interface haziness and a peripheral corneal steepening at 2 oʼclock ([Fig. 1 ]). The fellow eye demonstrated a central
crocodile shagreen corneal degeneration. At the time of first presentation, uncorrected near visual acuity in the affected right eye was hand motion. The anterior corneal astigmatism was
highly irregular and measured 11.2 diopters. Contact lens fitting to improve visual acuity was not possible due to contact lens intolerance. Therefore, we performed an excimer laser-assisted
penetrating repeat keratoplasty (8.5/8.6 mm) because of high irregular astigmatism.
Fig. 1 Anterior segment photograph and anterior segment optical coherence tomography (AS-OCT) of cases 1 (a, b ) and 2 (c, d ) after previous DALK.
a Preoperative examination with interface haziness, limbal relaxing incision (red arrow), and a corneal steepening at 2 oʼclock (circle). b AS-OCT demonstrates a corneal
steepening at 2 oʼclock (circle). The red arrow marks the scarred area of the limbal relaxing incision. Moderate haze formation is observed at the interface area. c Preoperative
findings show a clear corneal graft after previous DALK. d AS-OCT displays the margins of the corneal graft. Significant corneal steepening is not present.
Case 2
A 50-year-old male patient presented to our department due to high irregular astigmatism in the left eye 2 years after manual DALK. The surgical procedure was performed due to deep corneal
scarring after foreign body trauma. The corneal graft was clear without interface haziness and did not show any signs of graft rejection ([Fig. 2 ]). Visual acuity
was 6/120 in the affected eye. Tomography demonstrated an anterior corneal astigmatism of 8.9 diopters. There was no tolerance to contact lenses. In case of high irregular astigmatism, we
performed an excimer laser-assisted penetrating repeat keratoplasty (8.5/8.6 mm).
Fig. 2 Cross-sectional histologic specimen of the corneal explant after previous DALK. a Low power view of the corneal explant (Case 1) shows a second Descemetʼs membrane
(black arrow) that is not continuously apparent. More peripherally, it can only be identified dimly (arrowhead). The interface area does not reveal any vital keratocytes. Incidental
findings include a defect of Bowmanʼs layer with the start of epithelial invasion (star) (periodic acid-Schiff stain, original magnification 50×). b Higher power view of the
corneal explant (Case 1) to demonstrate the second Descemetʼs membrane, which is interrupted abruptly (black arrow). In addition, a disturbed collagen lamellae distribution is seen
peripherally (white arrow). The distance between both Descemetʼs membranes measured 25 µm (double-headed arrow) (periodic acid-Schiff stain, original magnification 100×). c Higher
power view of the corneal explant (Case 2) reveals an oblique tissue separation of the corneal stroma surrounded by macrophages and fibroblasts. Disruption of Bowmanʼs layer is also
visible (circle). No differentiation of donor and recipient stroma is possible (periodic acid-Schiff stain, original magnification 100×).
Light Microscopy
Case 1
Light microscopy (periodic acid-Schiff stain, PAS) of the central corneal button ([Fig. 2 a, b ]) revealed several fragments of a second PAS-positive DM of the
donor tissue in the peripheral and midperipheral regions of the cornea. The remaining recipient stroma measured about 25 µm in width (distance between both DMs) and showed partial structural
alterations with irregularities in collagen lamellae distribution. No keratocytes could be seen in the interface area. Incidental findings included disruption of Bowmanʼs layer with
beginning epithelial invasion into the corneal stroma. The corneal endothelium appeared regular.
Case 2
Light microscopic examination (PAS stain) of the corneal tissue ([Fig. 2 c ]) demonstrated a mainly intact epithelium that was centrally discontinuous at one
location with disruption of Bowmanʼs layer. The anterior corneal stroma revealed an oblique tissue separation surrounded by macrophages and fibroblasts. The remaining corneal stroma was
inconspicuous. No clear differentiation of donor and recipient stroma was possible. The DM was intact, and the endothelial cell layer appeared normal.
Transmission Electron Microscopy
Case 1
[Fig. 3 a, b ]: The interface area was identified by irregularities in the distribution of the collagen lamellae, although donor and host corneal stromas were
tightly attached without any dehiscence. The residual recipient stroma measured approximately 30 µm in width and was characterized by irregular arrangement of collagen lamellae, which were
interspersed with amorphous material and vacuolar inclusions, which also formed a band-like zone in the pre-Descemet area. Vital keratocytes were not observed in the recipient stroma, and
keratocyte density was also decreased in the posterior donor stroma compared with the anterior regions (not shown). The recipient lamella further showed a regular DM and degenerative
appearing endothelial cells. A second DM could not be observed by TEM.
Fig. 3 Transmission electron microscopy of the corneal explant after previous DALK (Case 1 – a, b and Case 2 – c, d ). a The recipient side demonstrates
degenerated endothelial cells and a normal Descemetʼs membrane. The interface area is characterized by irregularities of the collagen lamellae (black line). Vacuolated inclusions are
evident in the remaining recipient stroma (white arrow). A predescemetic vacuolated band is also observed (blue arrowhead). b The magnified section shows vacuolated inclusions in
the recipient cornea (white arrow), amorphous material (red arrow), and a vacuolated predescemetic band (blue arrowhead). c The interface area can be identified by irregularities
of the collagen lamellae (black line). The recipient side shows a normal Descemetʼs membrane. In the predescemetic area, the recipient stroma demonstrates degenerated keratocytes (white
star). d In another sectional area, degenerated keratocytes (white star) are also shown in the recipient stroma. A predescemetic vacuolated band is present (blue arrowhead).
Case 2
[Fig. 3 c, d ]: The interface area was less distinct than in Case 1, but could still be visualized by focal irregularities of the collagen lamellae. Few
degenerated keratocytes were seen in the recipient corneal stroma, which measured approximately 100 µm in width. A weak predescemetic vacuolar band-like alteration was also observed. The
recipient lamella further showed a regular DM and degenerated endothelial cells. Keratocyte density was again reduced in the posterior donor stroma compared with the anterior stromal regions
(not shown).
Discussion
The cellular responses and remodeling processes of the corneal interface after DALK are still largely unknown and are expected to be important for the morphologic and functional outcomes.
Currently, in vivo confocal microscopy is the most common method for examining the corneal interface after DALK. Generally, for quiescent keratocytes, only round or oval-shaped nuclei,
but not their cytoplasm, can be visualized. In contrast, activated keratocytes have a higher corneal light backscattering than quiescent keratocytes and their cellular processes are often
apparent [7 ]. After corneal injury, activated keratocytes function as repair cells and repopulate the acellular area by migration and proliferation. Through
excessive matrix production, they are able to remodel the extracellular matrix [7 ].
In vivo confocal microscopy studies have demonstrated an increased keratocyte reflectivity immediately after DALK [4 ], [6 ], [8 ], [9 ], [10 ]. In addition, differences in the morphology and reflectivity of keratocytes were observed, depending
on the depth of the recipient bed preparation. Keratocyte reflectivity was increased in pDALK and dDALK, but less pronounced in the latter. However, reflectivity decreased and keratocyte
morphology normalized in the interface area after a few months postoperatively [4 ], [6 ], [9 ].
Complete exposure of DM with removal of the predescemetic corneal stroma is an important prognostic factor in terms of postoperative visual outcome. If the residual thickness of the recipient
bed is less than 80 µm and is homogenous in its thickness, good visual outcomes may be achieved [5 ]. Compared to dDALK, postoperative visual recovery in pDALK takes
longer, and is reported to have 2 – 5 years of follow-up [11 ].
There are various methods for separating the stroma from the DM of the host cornea, such as viscodissection or pneumodissection, using a “big bubble” [12 ], [13 ].
Schlötzer-Schrehardt et al. [14 ] showed that the distance of stromal keratocytes to DM increases from the central (3.5 to 14 µm) to the peripheral (4.5 to 18 µm)
cornea.
Air injection into the corneal stroma of organ-cultured corneas (after DM stripping) resulted in a formation of a big bubble with a bubble wall composed of a stromal sheet varying in
thickness from 4.5 to 27.5 µm with keratocytes on the anterior surface. It was demonstrated that the separated stromal layer was thinnest in the central part of the cornea and became thicker
in the periphery, suggesting that dissection does not occur along a uniformly thick stromal layer. The authors assume that the variable thickness of the remaining stromal layer may be related
to the increasing distance of the keratocytes to DM from center to periphery. This indicates that the intrastromal cleavage after pneumodissection occurs along the posterior keratocytes due to
a lower biomechanical resistance [14 ]. Therefore, pneumodissection should not be considered as a DM-baring technique.
Generally, DALK should only be completed when the DM is fully exposed and not perforated. In case of scarring and uncertain depth during preparation, conversion to PKP should always be
chosen. Especially, excimer laser-assisted DALK allows an unproblematic conversion to PKP without disadvantages for the patient, provided the graft is also suitable for PKP [1 ], [2 ], [15 ]. In Case 2, conversion to PKP would have been the method of choice due to the deep stromal scarring,
but also due to the possibly limited visibility of the DM intraoperatively that resulted in a remaining host stromal thickness of 100 µm. The reduced number and degeneration of keratocytes and
extracellular changes observed in TEM ([Fig. 3 c, d ]) may result in a reduced long-term postoperative outcome. Therefore, the remaining host stroma should be
reduced to a minimum in order to obtain PKP-like visual results.
In addition to the preparation of the recipient bed with exposure of the DM, the method of donor preparation is also relevant. Using a full-thickness corneal graft (Descemet-on) was
associated with significantly higher haze formation in the interface area [10 ]. The histological section ([Fig. 2 a, b ]) of the corneal
button demonstrated a second DM, which was not continuously present and was only weakly pronounced peripherally. One explanation would be a failed preparation of the donor tissue with a tear
during the attempt to remove the DM and leaving it partially attached to the donor stroma.
On the other hand, increased keratocyte activation after DALK, which was confirmed by confocal studies, could come along with remodeling and degradation of DM and could be caused by increased
expression of matrix metalloproteinases [16 ]. Although enzymatic digestion is only hypothetic, it may provide another potential explanation for the faded DM.
Localization of the interface area was possible by in vivo confocal microscopy due to hyporeflective striae in the posterior stroma, representing microfolds, hyperreflective amorphous
material, and high contrast microdots [8 ], [10 ].
In our cases, TEM ([Fig. 3 ]) was clearly able to differentiate the recipient and donor stromas and thus identify the interface area based on irregularities in the
distribution of the collagen lamellae. However, no clear interface area could be detected by light microscopy, which was also reported by Favuzza et al. in a recent case series [17 ]. Differentiation between recipient and donor stromas based on light microscopy alone would not have been possible without indications like the presence of the second
DM. It is interesting that the group of Favuzza et al. could not identify any interface area by TEM. In contrast to our results, the keratocytes had a normal morphology and the posterior
stroma was arranged with regularly organized collagen lamellae without evidence of stromal irregularities [17 ]. Considering the fact that the group was able to
identify the exact location of the interface area by AS-OCT 3 weeks postoperatively, the absence of any evidence in the TEM even 2 years after DALK is surprising.
Abdelkader et al. investigated the corneal wound healing after DALK in rabbits. It was demonstrated that the keratocytes disappeared in the area adjacent to the wound edge and began to
repopulate 7 days postoperatively. Furthermore, highly reflective interface particles were detected, which could represent foreign cell debris or inflammatory cells [8 ].
In our cases, TEM revealed structural alterations in the interface region and recipient stroma, particularly irregularities in the collagen fiber arrangement, inclusion of amorphous and
vacuolar material, and formation of a vacuolar band-like zone anterior to DM. Vacuolar inclusions and amorphous material interspersed between collagen fibers were also described in confocal
studies [8 ], [10 ]. We hypothesize that these changes may indicate remnants of degenerated keratocytes due to increased apoptosis after
DALK, since increased keratocyte density is found directly predescemetally in healthy corneas [18 ]. TEM showed absence or degenerated keratocytes in the interface
area and in the recipient stroma as well as in the adjacent donor stroma. A confocal microscopy study by Feizi et al. revealed more distinct cellular changes in the donor tissue after DALK
compared to PKP. Keratocyte density after PKP was comparable to those of a normal cornea in the anterior, mid, and posterior stromas, whereas in DALK, keratocyte density was significantly
reduced in all three stromal layers. Also, the decrease in keratocyte density from anterior to posterior stroma was not seen in the DALK group [19 ].
The disturbed alignment of the collagen lamellae in TEM confirms the assumption of extracellular matrix remodeling processes after DALK.
In summary, up to now, the examination of the corneal interface has been performed mostly by confocal microscopy studies, while results from light and electron microscopy studies have been
described only sporadically [17 ]. Differentiation of the interface area could only be achieved with TEM and the findings seem to concur with results of previous
confocal microscopy studies. For a successful long-term postoperative result, the surgical method is of major importance, and it must be assumed that a stroma-stroma interaction is accompanied
with higher interface remodeling and haziness. Therefore, complete exposure of DM is an important prognostic factor regarding the postoperative visual outcome.
