In recent years the evolution of modern refractive surgery has focused attention on the architecture and biological properties of the cornea. In this issue of theBJO (p 437) Müller et al address the differential behaviour of the anterior and posterior stroma during corneal swelling and draw interesting conclusions about the factors maintaining corneal shape.

Transparency of the corneal stroma depends particularly on the degree of spatial order of its collagen fibrils which are narrow in diameter and closely packed in a regular array.1-8 The collagen fibrils themselves are weak scatterers, since their fibril diameter is less than the wavelength of light, and fibril refractive index is close to that of the ground substance. There is little variation in fibril diameter and separation between the anterior and posterior cornea.

The stromal fibrils are further organised into bundles, or lamellae, of which there are approximately 300 in the central cornea and 500 close to the limbus.9 The posterior lamellae course directly across the full width of the cornea without a break, having their origins in fibres which wind around the limbus at the corneoscleral junction10-12 or, according to Radner,9 have a pseudocircular organisation at the limbus, forming the ligamentum circulare corneae. On the basis ofx ray diffraction studies, about 49% of the stromal lamellae are preferentially aligned orthogonally, along the vertical and horizontal meridians, while about 66% lie within a 45° sector.1112 Fibrils within a lamella are in parallel array, except where branching of lamellae occurs. Branching in the horizontal plane occurs throughout the stroma, whereas anteroposterior branching is found only in the anterior third.13

The anterior and posterior stroma differ in specific ways. In general the posterior stroma is more ordered,14 more hydrated,15 more easily swollen, and has a lower refractive index16 than the anterior stroma. The posterior lamellae are also wider and thicker (100–200 μm wide and 1.0–2.5 μm thick) than the anterior (0.5–30 μm wide and 0.2–1.2 μm thick).13 There are also differences in keratocyte morphology.17 It has long been established that the posterior lamellae of the human corneal stroma are arranged parallel to the plane of the corneal curvature13 and this feature is recognised to facilitate dissection in lamellar corneal grafting.1819 Dissection of the cornea is, however, not resistance free, suggesting that there are elements which bind the collagen lamellae together.20 Part of this resistance is likely be due to attachments between the collagen fibrils on the one hand and other matrix proteins such as the proteoglycans2122 or keratoepithelin.23

In the anterior stroma, an additional contribution is made by the marked anteroposterior lamellar interweave which has been recognised to be a feature of the corneal architecture since the early part of the century.91324-33 Here, lamellae can be shown to pass obliquely from one layer to another, sometimes passing across several lamellae to reach their destination.13 It is likely that such obliquely disposed lamellae have their peripheral origins in the limbus, although this specific question has never been explored directly. A proportion of the anterior lamellae are known to be inserted directly into Bowman's layer and it has been suggested that the latter contribute to the formation of the anterior corneal mosaic, a normal architectural feature seen at the corneal surface.3035-38 The anterior corneal mosaic is visible in all normal corneas as a broad polygonal pattern which can be observed after instillation of fluorescein, simply by exerting pressure on the cornea through the closed lids, and observing the fluorescein distribution when the eyes open. This polygonal pattern can be regarded as the most superficial manifestation of a more complex, three dimensional “chicken wire” arrangement of the anterior stromal lamellae.30

In this issue, Müller et al elaborate at ultrastructural level, an older, light microscopic observation, that human anterior stroma swells considerably less than the posterior stroma, when corneas are immersed for a prolonged period in saline.39-46 In non-nutrient media, at room temperature, where there are no cellular barriers, and no viable cells capable of deswelling the stroma, stromal swelling is due almost entirely to the gel pressure exerted by the stromal proteoglycans, acting as a polyelectrolyte gel.47 It is the high, negative charge of the glycosaminoglycan (GAG) components of the proteoglycans, that is responsible for this property. Müller et alclaim that the anterior stroma, 100–120 μm deep to Bowman's layer, does not swell perceptibly when the cornea is immersed in water or saline for prolonged periods and that swelling is confined to the posterior stroma. This is a remarkable observation that implies that the anterior stroma has special features which constrain swelling in these conditions, despite the presence of negatively charged proteoglycans here, as in the posterior stroma. These observations are important and need to be confirmed by morphometric measurements of fibril number density (fibril number per unit area) in the respective zones, with special attention to the presence or absence of stromal “lakes”.

There are a number of factors that could explain the findings of Müller et al. As noted above, the morphology of the anterior and posterior stroma differs considerably. Müller et al suggest that the anterior stromal interweave is the chief architectural factor determining the differential swelling behaviour of the stroma. They also suggest that it is responsible for the structural stability of this region of the cornea, a feature which is of importance to refractive surgery and possibly in such conditions as keratoconus. However, as these and other authors have observed, additional factors may contribute to the differential swelling. The GAGs of the corneal stroma are keratan sulphate (a component, for instance, of the proteoglycan lumican), dermatan sulphate (DS), and chondroitin sulphate (CS) (components of the small proteoglycan CS/DS proteoglycan, decorin). Keratan sulphate makes up about 50% of the corneal GAGs. In bovine corneal stroma, the keratan sulphate/chondroitin-4-sulphate ratio is higher posteriorly than anteriorly.4244 If this is the case for human cornea, then since keratan sulphate has a higher water affinity than chondroitin-4-sulphate, this could explain, in part, the greater degree of posterior stromal swelling on immersion. Another factor, which should be kept in mind, is the possibility of a differential leaching of GAGs from the stroma during prolonged immersion. Although only about 1% of keratan sulphate is lost from corneas held in closed culture48 a significant loss of proteoglycans from swollen corneas has been recorded by others4549 with a preferential loss of keratan sulphate from oedematous rabbit corneas.50 Differential loss has not been studied, but a greater loss of GAGs from the anterior stroma could reduce its “swellability”.

This behaviour of human anterior stroma is reminiscent of that of the stroma of the cartilaginous fishes (Chondricthyses, which includes the subclass of elasmobranchs). The stromal lamellae in such fish are well defined and run in the plane of the cornea, but are crossed at right angles by anteroposterior bundles of “sutural” fibres or complexes, which connect the basal lamina of the corneal epithelium to a posterior collagenous layer, resembling Descemet's layer in location but not structure.51 The sutural fibres were first described by Ranvier52 and subsequently by Payrauet al,53 Goldmann and Benedek,54 and Faure.55 The corneas of the cartilaginous fish swell little when immersed in water, and retain their transparency56 apparently in the absence of a functional endothelial layer.515758 It has been suggested that the sutural fibres provide a restraining action on corneal swelling, possibly assisted by an interaction between stromal collagen and stromal matrix materials,54 which are abundant, for instance, in the dogfish cornea. It appears that the anterior interweave of the stromal lamellae of the human cornea and, possibly, differences in proteoglycan composition and attachment may play a similar part to that of the sutural fibres in the cartilaginous fish, whose lamellae show little or no anteroposterior interweave.

The anterior stromal interweave has other structural implications for the cornea. It can be conceived that while the limbus to limbus arrangement of the posterior lamellae offers a singular advantage with respect to strength, the interweave of the anterior lamellae, and the insertion of lamellae into Bowman's layer, offers opportunities to confer a variable shape to the anterior corneal surface. Although the insertions of lamellae into Bowman's layer might seem to offer less structural strength than the limbus to limbus arrangement of the posterior stroma, loss of strength would be minimised if anterior insertions extended from the limbus to Bowman's layer, beyond the corneal centre. This might also afford better opportunities to determine shape. Since corneal shape is to some extent hereditable, the inference would be that the anterior obliquities are under genetic control and regulated by proteins whose spatiotemporal distribution during development determine corneal shape. It is relevant that the developmental origin of the anterior third of the corneal stroma is thought to differ from that of the posterior.25

Müller et al suggest that the structural stability of the anterior stroma under conditions of extreme hydration imply an important role for this zone in the maintenance of corneal curvature and that this stability is determined by the tight interweave of the stromal lamellae here. It seems a reasonable proposition that the interweave is important in maintaining shape and it seems likely too that is a determinant of shape, probably by distributing tension over the corneal surface in a manner which could not be achieved by an interlimbal arrangement alone.

One final implication of the human anterior stromal interweave should be considered. It is generally accepted that anterior stromal keratocytes die shortly after the induction of a corneal abrasion. It has reasonably been proposed, by Wilson,60 that this is due to a FAS-FAS ligand mechanism, in response to IL-1 release from damaged epithelium. However, an alternative explanation could be advanced, that corneal abrasion, by exposing the corneal stroma to the tears, tends to cause stromal swelling. If gel swelling of the anterior stroma is restricted by the stromal interweave, then a rise in anterior stromal hydrostatic pressure would result. We may at least ask ourselves the question, could keratocyte loss be caused by such a rise in pressure, do the keratocytes die because they are “strangled” by the stromal interweave? This could also explain the preferential loss of anterior stromal keratocytes which is said to occur in bullous keratopathy.

What influence does the anterior stromal architecture have on refractive procedures? Müller et al caution that removal of this critical, stable zone of the stroma during photorefractive keratectomy (PRK), could lead to later optical problems. This may not be the case for most PRK ablations, since the depth of ablation, say 70 μm deep to the surface of Bowman's layer, may leave untouched a 50–60 μm zone of the interwoven, anterior region of the stroma, capable of providing some structural rigidity to the newly sculpted zone. As noted by Müller et al, since the combined thickness of the epithelium and Bowman's layer together, is about 60 μm, a LASIK flap of 160–180 μm will just encompass the interwoven anterior stromal layer (100–120 μm thick). A deeper plane could cut into the interlimbal lamellae of the posterior stroma and, potentially, interfere with the stability of the procedure, much as Müller et al propose. It may be noted, in passing, that Munoz et al59 devised a method for dealing with wrinkling of the LASIK flap, which involves “rehydration” of the flap with distilled water. It must be supposed that the distilled water swells and stretches a hydratable posterior lamella of the flap to achieve this effect.

In summary, Müller et al have drawn our attention to important structural and functional features of the cornea which are not only important in maintaining corneal curvature, but may also play an important part in determining corneal shape. The realisation of this may have far reaching consequences for our understanding of the corneal response to injury and of the biological response to refractive corneal procedures. It is clearly an area that deserves further attention.