BritishJournal ofOphthalmology, 1990,74,519-522

Staining of the conjunctiva and conjunctival tearfilm

Joseph A Eliason, David M Maurice

AbstractBoth the distribution of tear fluid over theconjunctiva and any injury to the conjunctivalepithelium can be made visible by instilling afluorescent solution into the eye and observingwith an appropriate combination of excitationand barrier filters. Sulphorhodamine B, whichhas an orange fluorescence that can be sepa-rated from the green natural fluorescence ofthe ocular tissues, gives a greater contrast thanfluorescein. The tear film is seen to cover thesurface of the conjunctiva and to be concen-trated in its folds. Small circular areas of thintear film appear transiently in the neighbour-hood of the limbus after a blink. Occasionalcells stain on the normal conjunctiva, partic-ularly in the interpalpebral area. The density ofthe staining increases in dry eye conditions.Conjunctival trauma is sensitively revealed bythe method, and its healing can be followed.Hard contact lenses are seen to traumatisecontinually the inferior limbal conjunctiva insymptomless wearers.

Department ofOphthalmology, StanfordUniversity MedicalCenter, Stanford, USAJ A EliasonD M MauriceCorrespondence to:D Maurice, Department ofOphthalmology A157,Stanford University MedicalCenter, Stanford, CA 94305,USA.Accepted for publication15 March 1990

The instillation of a drop of fluorescein is a

common method of detecting injury to thecorneal epithelium. Before damaged cells arerevealed by the stain, the precorneal tear film iscoloured by the dye, and it is possible to examineit for normal variations in uniformity or for theappearance ofdark patches when the layer breaksup. The conjunctival cells and the precon-junctival tear film may be of equal physiologicaland pathological interest to that of the cornealepithelium, but no equivalent method of stainingthem has been available. Rose Bengal isemployed as a stain for conjunctival damage, butit provides poor contrast as usually applied, andis uncomfortable when instilled in the eye.

There are three reasons why staining withfluorescein has not been found valuable over theconjunctival surface:

(1) The use ofan inefficient optical filter system.Generally, a cobalt filter has been placed in frontof the slit-lamp light and the eye has beenobserved under the microscope without theintervention of a barrier filter. This is adequatefor the observation over the dark background ofthe cornea, but the scattering ofthe incident bluelight by the sclera hides the weaker fluorescentlight emitted by the dye.

(2) The background fluorescence of the tissue.Even with an efficient pair of excitation andbarrier filters the blue incident light will excite a

greenish fluorescence in the conjunctiva andsclera, which is seen as a general glow thatreduces the contrast of the fluorescein.

(3) Penetration of fluorescein. Because of theslight lipid solubility of the dye, it can stain theconjunctiva tissue appreciably, and this further

reduces the contrast between the surface and thebackground.

It was thought possible that problem (3) couldbe overcome by the use ofa more hydrophilic dyesuch as carboxyfluorescein, but it was decided toevade problem (2) as well by using the fluoro-phore, sulphorhodamine B, which is not onlyvery hydrophilic but can be excited by wave-lengths too long to lead to visible fluorescence inthe tissues. When observed with an appropriatepair of colour filters, it allowed both the tear filmand devitalised confunctival cells to be readilyobserved with high contrast against the tissuebackground.The purpose of this report is to emphasise the

advantages ofan appropriate pair of colour filterswhen fluorescent dyes are used to examine theconjunctival surface, and to illustrate theadditional benefit of using a dye with largerwavelength fluorescent characteristics thanfluorescein. In addition, we note some hithertoundescribed physiological and anatomical prop-erties of the conjunctival tear film and surveysome clinical conditions in which the techniquemight prove to have value. (These observationshave already been briefly noted in the abstracts ofthe First Meeting of the International Society ofDacriology, Budapest 1987, and ARVO 1988.)

Materials and methodsSulphorhodamine B (molecular weight 559) is similar inchemical structure and properties to fluorescein but is lesslipid soluble, with an octanol/water partition coefficient of0-0033 compared with 0-61 for fluorescein.' Its fluores-cence is orange and is most readily excited by green light,the peak absorption and emission wavelengths being 556and 572 nm. It is usually supplied as an acid and must bedissolved in alkali, being soluble to the extent of about20 g/l at pH 7. The toxicity of the compound appears to bevery low. Doses of up to 1 g/kg could be injectedintraperitoneally in the mouse without causing death, and50 [d of the saturated solution injected into the stroma of arabbit cornea had as little effect as an equal volume ofsaline.The subjects were patients, residents, and staff of the

ophthalmology clinic. Informed consent was obtainedafter the nature of the procedure had been fully explained.A 0-5% solution of the dye at pH 7, made 300 mM withsucrose, was used to stain the tear film and a small dropwas applied to the tarsal conjunctiva with a woodenapplicator. The solution did not cause discomfort in theeye of any volunteer or patient. It is not important,however, to adhere exactly to this dye formulation.The eye was observed with a Haag-Streit 900 slit-lamp

using a broad patch of illumination. A band-pass inter-ference filter of 520-550 nm, such as is used as a barrierfilter for fluorescein, was inserted in the illuminating lightpath, and gelatin absorption filters (Kodak No 22)transmitting 100% of the light at 600 nm were placed inthe eyepieces. This combination of filters passes less than0-01% of light at the cross-over point of their absorptioncurves, around 570 nm. Photographs were taken with aZeiss photoflash slit-lamp on 1600 ASA colour film withthe same combination of filters. Black-and-white film ofsimilar speed can also be used.

519

on Decem

ber 2, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.74.9.519 on 1 Septem

ber 1990. Dow

nloaded from

Eliason, Maurice520

Figure I Staining of tearfilm with sulphorhodamineB in a normal subject. ((A)Pattern over conjunctivaimmediately after a blink.(B) Same area ofconjunctiva a few secondslater. Note the enhancementof lines and patches withtime.

Figure IA Figure IB

Figure 2 Disappearanceofconjunctival pattern withgentle pressure from atransparent plastic plate(arrow).

Results

NORMAL TEAR FILMIn the absence of the dye the tissues of the eyeappeared black through the microscope, makingfocusing on the ocular surface difficult. Pin-gueculae gave rise to a natural fluorescence whichwas absent in normal sclera or cornea.

After the dye was instilled the precorneal tearfilm and the lacrimal strips shone as brilliantly asthey do with fluorescein under blue illumination.Furthermore, the conjunctiva was seen, in thefirst moments after instillation, to be coveredwith a more or less uniform tear film. Shortlythereafter a system of lines and patches appearedwhich became more prominent during the firstseconds immediately after opening the eye, whilethe fluorescence of the tear film covering the areabetween them became weaker (Figs 1A, B).These seem to be a result of accumulations of

fluid rather than tissue staining, since theydisappeared after thorough flushing with salineor on pressing the conjunctiva under a clearplastic plate (Fig 2). Many features of thepatterns of the lines remained constant over along period, and they almost certainly corres-pond to folds in the conjunctiva (Figs 3A, B).The wider patches may represent layers ofmucusor broad depressions in the conjunctival surface.The thickness of the mucus layer on the con-junctiva has been reported to vary over thesurface.2Some transient phenomena occur after a blink.

The tear film over the entire cornea and con-junctiva appeared to flow upwards immediatelyafter the opening of the eye. This movementlasted for about 3 seconds. Often circular darkpatches about 1 mm in diameter were observed inthe tear film over the conjunctiva and cornea,most frequently near the limbus (Fig 4). Theyappeared immediately after a blink, flowedupward with the tear film, and vanished in asecond or two by constricting to smaller dia-meters. The origin of these spots has not beenexamined, but it seems most likely that theywould correspond to an area where the aqueouslayer is thinner rather than where it is diluted.

CONJUNCTIVAL STAININGIf the dye was removed from the tear film bythorough flushing with saline, or if it was allowedtime to be washed out in the normal tear secre-tions, the background that remained was eitherdark or showed a faint orange glow from dye that

Figure 3 Staining withsulphorhodamine B in anormal subject. (A) Patternofconjunctival lines. (B)Same area ofconjunctivaafter an interval ofa year.Arrows indicate some of thesimilarities.

Figure 3A Figure 3B

on Decem

ber 2, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.74.9.519 on 1 Septem

ber 1990. Dow

nloaded from

Staining ofthe conjunctiva and conjunctival tearfilm

Figure 4 Dark patches (arrows) noticed here on both sidesofthe limbus, which were observed to move with the tearfilm.

Figure 5 Conjunctivalstaining along the limbus inthe interpalpebral zone in anormal eye. Also noteprominent staining of thetear meniscus along thelower lid margin.

had penetrated into the tissue. In the normal eyebright staining spots were seen, occasionally onthe cornea but more frequently over the con-junctiva, particularly the interpalpebral zone(Fig 5). Sometimes small patches ofstaining wereobserved. The spots retained sharp outlines andcould be observed for hours, though usually theydisappeared by the next day and were not

constant on retaining the tissue. They probablycorrespond to individual epithelial cells that areeither damaged or in a permeable conditionbefore desquamation. Patches of staining alsoappeared where the tarsal or bulbar conjunctivahad been touched by the applicator. In one eye abrilliantly staining zone of conjunctival damagewas created by the rotation ofthe wooden end ofasterile Q-tip, and its recovery healing could befollowed by retaining at intervals (Fig 6A, B). Inthis case a portion of the initially applied stainwas still present after 24 hours.

In each one offour symptomless gas permeablecontact-lens wearers notable staining in the formof a confluent band superimposed on a fine linewas found on the conjunctival side of the inferiorlimbus (Fig 7).

COMPARISON WITH FLUORESCEIN AND ROSEBENGALIn 12 patients with Sj0gren's syndrome con-junctival staining with sulphorhodamine B wasfound to be more evident than that detected withfluorescein (Figs 8A, B). Corneal staining wassimilar with both dyes. This subjective impres-sion that sulphorhodamine B staining was moresensitive than that with the other dyes was given aquantitative test. Observations were carried outon 14 eyes of seven normal subjects with 1%fluorescein, 0 5% rose Bengal, and 0 5% sul-phorhodamuine B. The blue and green excitationfilters provided with the Haag-Streit 900 slit-lamp were used to observe the fluorescein androse Bengal dyes, and the excitation and barrierfilters described above were used for the sul-phorhodamine. While this arrangement shouldprovide better contrast for viewing the sul-phorhodamine, the recognition ofpunctate stain-ing spots over the conjunctiva with fluoresceinwas judged similar with or without the use of abarrier filter. Punctate staining was quantitatedby the grading scheme described by van Bijster-veld3 using four areas: cornea, and nasal, tem-poral and inferior conjunctiva. Each area wasgraded on a 0 to 3 scale: 0=no punctate staining;1=small (that is, readily countable) number ofpunctate staining spots; 2=numerous (that is,not countable) spots; and 3=staining that wasconfluent or approaching confluence. The scoresfor each eye were added to reach a total of0 to 12.

Figure 6: SulphorhodamineB staining of the conjunctivain a normal subject. (A)Area ofconjunctivaldamage stained immediatelyafter its production. (B)Restained 24 hours later,showing almost completehealing ofthe defect.

Figure 6A

521

Figure 6B

on Decem

ber 2, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.74.9.519 on 1 Septem

ber 1990. Dow

nloaded from

Eliason, Maurice

Figure 7 Bright band ofstaining along the inferiorlimbus in a rigid contact lens

wearer. Also note faint edgeof lens riding on the cornea(arrow).

Mean scores for the dyes were: fluorescein1 0, rose Bengal [-2, and sulphorhodamine 2-7.Evaluated with the non-parametric Wilcoxonsigned rank test, the difference between thefluorescein and rose Bengal scores was significantonly at the 5% level, while the differencesbetween either fluorescein or rose Bengal andsulphorhodamine were significant at the 1%level.

DiscussionFluorescein staining of the cornea or the precor-neal tear film can be readily observed under awhite light or with a cobalt filter, but the dye isscarcely visible on the conjunctival surface. Theuse of an efficient pair of excitation and barrierfilters leads to a considerable increase in contrastof the dye and allows conjunctival lesions to beidentified. However, the blood vessels of theconjunctiva show up as a system of black lineswhich is disturbing, and background fluores-cence is appreciably greater following washout ofthe dye, so that contrast is poorer and thedetection of small lesions is difficult. The linesare not visible when sulphorhodamine B replacesfluorescein, and the contrast over the conjunctivais much improved. The colouration with roseBengal observed in white light seemed to be theweakest of all, but it was more visible whenexcited with green light.From our comparison of the three dyes we

consider that sulphorhodamine B is the mostcapable ofrevealing subtle surface abnormalities.Although fluorescein under our operating condi-tions resulted in an image easier to photograph,this is not an important consideration for visualobservations. Punctate staining with sulphor-hodamine B could be detected in all eyes con-sidered normal and asymptomatic which weexamined. An increase in the sensitivity ofdetection of ocular surface abnormalities mayresult in new diagnostic insights. Thus theevidence of limbal epithelial damage in asymp-tomatic contact lens wearers we believe to beoriginal. It is probable that this considerableepithelial injury occurs when the edge of the lensis driven down on to the limbus on blinking.Damage to the important repopulating limbalepithelial cells may explain the poor surfacehealing behaviour observed in long-term lenswearers.

Further experience is required to determinewhether this increase in sensitivity will lead to animprovement in diagnosis sufficient to warrantthe general use of sulphorhodamine B. Animportant prerequisite is a method of readilyinserting colour filters in the lamp and themicroscope optical pathways. The system wehave employed is too slow and awkward forregular use.

This work was supported by EY 04863 and Research to PreventBlindness, Inc.

Figure 8i

Figure 8B

Figure 8 (A) Conjunctival staining in a patient with primary Sj0gren's syndrome by meansoffluorescein photographed with excitation and barrierfilters. (B) Same location withsulphorhodamine B. Note large area ofconfluent staining in the interpalpebral fissure area notevident with fluorescein and absence ofdistracting blood vessels.

1 Araie M, Maurice D. The rate of diffusion of fluorophoresthrough the corneal epithelium and stroma. Exp Eye Res1987; 44: 73-87.

2 Nichols BA, Chiappino ML, Dawson CR. Demonstration of themucous layer of the tear films by electron microscopy. InvestOphthalmol VisSci 1985; 26: 464-73.

3 van Bijsterveld OP. Diagnostic tests in the sicca syndrome. ArchOphthalmol 1969; 82: 10-4.

522

on Decem

ber 2, 2020 by guest. Protected by copyright.

http://bjo.bmj.com

/B

r J Ophthalm

ol: first published as 10.1136/bjo.74.9.519 on 1 Septem

ber 1990. Dow

nloaded from