Retinopathy - Archives of Disease in Childhood · Archives ofDisease in Childhood, 1988, 63, 1151-1167 Personal paper Retinopathy of prematurity: clinical implications of retinal

Archives of Disease in Childhood, 1988, 63, 1151-1167

Personal paper

Retinopathy of prematurity: clinical implications ofretinal developmentF L KRETZER AND H M HITTNER

Baylor College of Medicine, Houston, Texas, USA

Retinopathy of prematurity (ROP) is a potentiallyblinding ocular disorder that is unique to thepreterm infant. The current epidemic affects thesmallest and illest preterm survivors of modernneonatal intensive care units. With the prevention ofneonatal asphyxia, judicious curtailment of oxygen,and adequate doses of vitamin E, ROP has becomea disorder that is generally restricted to infantsweighing less than 1001 g at birth. The size of thesurviving population of infants with ROP dwarfs allprevious predictions of absolute numbers becausesurvival rates of infants weighing 501-750 g at birthare now approaching 60%, and those of infantsweighing 751-1000 g are now approaching 90%.

The total incidence of ROP in infants weighing501-750 g is almost 100% with severe ROPdeveloping in about 30%. The total incidence ofROP in infants weighing 751-1000 g is almost 80%with severe ROP developing in about 10%. Whetherthe future total incidence of ROP and that of severeROP will change is dependent on the unknowneffects of the many new developments in the care ofpremature infants.

NORMAL RETINAL DEVELOPMENTThe immature retina has two blood supplies, thechoroidal vessels and the inner retinal vessels (fig 1).The choroidal vasculature develops early, lies on the

Oc

CetaOptic disc

O = Canalising region (site of future shunt)

i = Specialised region (site of future macula)|

Fig 1 Four interrelated and concomitant processes (arrows) that regulate normal retinal development. Note thinning ofretina towards ora serrata, region offuture shunt whiere ROP is first seen clinically, and region offuture macula that willdetermine ultimate visual acuity.

1151

L-

Protected by copyright.

on Septem

ber 30, 2020 by guest.http://adc.bm

j.com/

Arch D

is Child: first published as 10.1136/adc.63.10_S

pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/

1152 Kretzer and Hittner

outer surface of the retina, and is the sole supplier ofnourishment to the thin, undifferentiated retina.The inner retinal vasculature develops late, lieswithin the inner surface of the retina, and nourishesthe inner portions of the thick, maturing retina. Thisinner retinal vasculature develops from a peripheralmigration of spindle cells from the optic disc.In utero, spindle cells in the thin, peripheral,avascular retina reside in a comparatively hypoxicenvironment.

PATHOLOGICAL EVENTS OF NEOVASCULARISATIONAND TRACTIONAfter premature birth, spindle cells in the thin,peripheral, avascular retina are stressed by acomparatively hyperoxic environment. This stimulusfor abnormal retinal development occurs becauseoxygen diffuses freely across the retina from thechoroidal vasculature that cannot vasoconstrict.Spindle cells stressed by free radical damage stopmigrating peripherally, stop forming inner retinalvessels by the process of canalisation, and begin tosynthesise and secrete angiogenic factors. For abouteight to 10 weeks these angiogenic factors induceabnormal inner retinal neovascularisation at theboundary between the vascular and avascular retina(fig 1). This boundary becomes a shunt that containsa large volume of rapidly flowing blood.1" Afterthis two month period myofibroblasts begin todifferentiate from a stem cell line in the shunt, theyform contractile sheets that invade the vitreous forabout four months, and they may cause tractionleading to retinal detachment.1 34 This entirecascade of events can be interrupted and regressspontaneously, or can continue to progress relent-lessly. In either case, the overall enlargement of theeye continues. This pathogenesis is different fromother neovascularisations of the mature eye, such asthose associated with diabetes and sickle cell disease.

Retinal development

There are four interrelated and concomitantprocesses that are important to an understanding ofROP (fig 1). These consist of a central/peripheralvector of retinal development that is coupled simul-taneously with an inner/outer vector of retinaldifferentiation, a remodelling of the macular region,and overall ocular growth.

CENTRAIJPERIPHERAL AND INNER/OUTER VECTORSRetinal development into the outer retina,connectors, and inner retina begins centrally at theoptic disc and advances peripherally to theora serrata. With respect to ROP there are twoimportant central/peripheral developmental pro-

cesses: photoreceptors in the outer retina andspindle cells in the inner retina (fig 1).

Photoreceptors begin to differentiate from theouter neuroblastic retina and advance as a gradientof maturation (figs 1 and 2). There are at least twostimuli for the normal formation of inner retinalvessels. One stimulus is the thickening of thematuring retina that displaces the inner retinafurther and further from the choroidal vessels.Another stimulus is that as photoreceptors maturetheir metabolic rate increases, and they utilise moreand more oxygen that diffuses from the nonvaso-constricting choroidal vasculature. Maturing photo-receptors secrete interstitial retinol binding proteininto the space between the retinal pigment epitheliumand the neural retina. This protein appears roundthe optic disc as early as 20 weeks' gestation andreaches the ora serrata at 29 weeks' gestation (atotal transit time of nine weeks).5 Its presence in thesubretinal space is critical to the potential retinaluptake of fat soluble vitamin E, and its extentestablishes the area of retina that can be protectedby vitamin E.6 After premature birth oxygen candiffuse easily from the non-vasoconstricting choroidalvasculature, can cause physiological vasoconstrictionof the inner retinal vasculature (but not endothelialnecrosis and retraction), and can create an abnormalhyperoxic environment for spindle cells. Interstitialretinol binding protein carries vitamin E (only in thepreterm infant) into the connectors (Muller cells),and the connectors carry vitamin E across the retinato spindle cells.

Spindle cells migrate in the inner retina from theoptic disc to the ora serrata (figs 1 and 2). Spindlecells appear round the optic disc at 16 weeks'gestation and reach the ora serrata at 29 weeks'gestation (a total transit time of 13 weeks). After29 weeks' gestation the spindle cell apron continuallydecreases in its linear extent so that the spindle cellapron has completely disappeared and the innerretinal vasculature is completely formed at40 weeks' gestation. In the smallest infants theproximal portion of the spindle cell apron isintermingled with the formed vessels, and a fewspindle cells may even remain in the adventitia ofthe hyaloid artery at the optic disc; in larger infantsthe spindle cell apron is not intermingled with theformed vessels, and spindle cells do not remain atthe optic disc. A small region of the spindle cellapron anterior to the last formed vessels continuesthe inner retinal vasculature by canalisation.The transretinal geometry between the slowly

migrating spindle cells (13 weeks transit time) andthe faster developing photoreceptors (nine weekstransit time) that are secreting interstitial retinolbinding protein changes with gestational age.7 In


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/

Retinopathy of prematurity: clinical implications of retinal development 1153D _.s:tHw Ia,X~~ p ^ .t w

| e ; ~

t lat.*er .s.n I N\ MtS~~~~.~ ~; : E 2~~~~~~~~~~~|~~~~~~~~~~~~~~~~- -~ s&JQZhSZ9 r w istE~~*~:*-

''_B6X.Wt._'" , -9, . t , b t >~~~~~~| X=t .S S 2~~~~~~~~~| s, w w a] b~~~~~~~~~

1-1 gZb , l_ *U T E~~i S ~~~~~~~~~rj**.*-Aft.s;.

.900- il:.--.

c 'IFiFullm


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


infants weighing less than 1001 g at birth (27 weeks'gestation or less), most of the spindle cell apron istransretinal to photoreceptors that are not secretinginterstitial retinol binding protein, and thereforethese spindle cells are less likely to receive adequateprotection from vitamin E. Thus even if vitamin E isgiven and oxygen is judiciously curtailed ROP maystill be severe in the iller infants weighing less than1001 g at birth. This morphological fact explains theclinical reality that vitamin E supplementation toeliminate plasma deficiency, even supplementationto pharmacological plasma concentrations, cannotprotect all surviving infants from ROP.

In infants weighing 1001-1500 g at birth(28-32 weeks' gestation), most of the spindle cellapron is transretinal to photoreceptors that are

4e

.X 39-0

3 38-U-

+ 37 -cna- 36-c

o 35-

4-

3-33-0.a-Q

secreting interstitial retinol binding protein, andtherefore these spindle cells can receive adequateprotection from vitamin E. Thus if vitamin E isgiven and oxygen is judiciously curtailed, ROP isusually not severe in these infants. Without vitamin E,however, ROP may be severe in the illest of theinfants weighing 1001-1500 g at birth. Thus inanalysing the results of trials of vitamin E forprotection against ROP, a multivariate analysis withbirth weight as one variable must be used ratherthan a univariate analysis. In infants weighing morethan 1500 g at birth (more than 32 weeks' gestation),fewer and fewer spindle cells remain within theretina. Thus whether or not vitamin E is given, but ifoxygen is judiciously curtailed, severe ROP does notusually develop.

Zones (ICROP) at - 10weeks

< I >!p II1000'

900

%800

3 700

LI 600

500

400

N0

3

IDU)

0m-o

0-9'a

tou

1.0 1.5 2.0 2.5 3.0Horizontal vascularised retinal length

(cm) at -10 weeksFig 3 Extent ofthe vascularised retina (horizontal vascularised retinal length (cm) at about 10 weeks) and zones(international classification ofROP at about 10 weeks ofage) is associated with birth weight (g), appropriate gestationalage+about 10 weeks oflife, and zones at about 10 weeks ofage. Horizontal lines denote area ofobligate zone 1 ROP.Vertical lines denote area ofobligate zone II ROP. Combined horizontal and vertical lines denote area ofzone I or IIROP(area ofbiological variability).


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/

Retinopathy of prematurity: clinical implications of retinal development 1155

The extent of inner retinal vascularisation isdirectly associated with gestational age and birthweight (fig 3). The clinical importance of thisassociation is that if severe ROP develops eight to10 weeks after birth in infants appropriate forgestational age, those weighing 600 g or less willhave ROP in zone I and those weighing 801 g ormore will have ROP in zone II. Those infantsweighing 601-800 g may have ROP in either zone Ior zone II.8 This reflects the normal biologicalvariability seen in all developmental processes. Thusthe inner retinal vasculature establishes the area ofretina that equates to the zones in the internationalclassification of ROP 9 10 that does not secreteangiogenic factors,4 and that may be rescued bycryotherapy to the peripheral retina where thespindle cells reside. Cryotherapy entails placing acold probe on the external surface of the eye todestroy all the living cells between the externalsurface and the vitreous. Retinal development alongthe central/peripheral vector stops after cryotherapy,but further retinal development associated with theinner/outer vector continues.

MACULAR REMODELLINGThe macula is a specialised region of the posteriorretina that is responsible for all fine visual acuity(fig 1). It is an avascular region in which the innerretina is displaced circumferentially 3600 so that lightcan impinge perpendicularly on the perfectly paralleldiscs of the cone photoreceptors. The commitmentto macular development and its subsequentremodelling must be triggered by an interplaybetween the inner/outer vector. This region isincompletely developed until three months afterbirth in an infant born at full term. Clinically themacula appears immature before this time-that is,it does not have a dimpled appearance.The distance between the temporal rim of the

optic disc and the centre of the macula is constant inall normally developing or developed human eyes.With the development of severe ROP the distancebetween the temporal rim of the optic disc and thecentre of the macula may be increased by dragging,whether the ROP regresses spontaneously or istreated successfully by cryotherapy.

OCULAR GROWTHAs the retina develops along the central/peripheraland inner/outer vectors with macular remodelling,the overall diameter of the globe increases rapidlywith only 15% of the increase in the diameter of theglobe occurring after40weeks'gestation (figs 1 and 4).Unless phthisis (ocular death with shrinkage) occurs,the globe continues to enlarge after the developmentof severe ROP, with or without surgical intervention.

Whether ROP is regressing or progressing, thisconcept of ocular growth is important when retinaldetachment caused by severe traction (fromsubstantial invasion by myofibroblasts) occurs.

Neovascularisation (spindle cells)

INITIATIONAfter preterm birth normally migrating and canal-ising spindle cells can become stressed by freeradicals. The probability that this stress to spindlecells stimulates the development of severe ROP isgreatest in the smallest and illest preterm infants.The stress is seen ultrastructurally as early as thefourth day of life as a formation of extensive gapjunctions between adjacent spindle cells uniformlythroughout the apron (fig Sa, Sc). Shortly afterwardsthese gap junction linked spindle cells contain anextensive volume of rough endoplasmic reticulum(fig Sb, Sd). This indicates increased communicationbetween cells (gap junctions), and increased proteinsynthesis and secretion (rough endoplasmicreticulum). Increased amounts of angiogenic factorshave been detected only from homogenates of theperipheral retina when gap junction linked spindlecells are present.4 7 This state exists for about eightto 10 weeks. 1

After eight to 10 weeks of synthesis and secretionof angiogenic factors, extensive gap junctionsbetween adjacent spindle cells disappear uniformlythroughout the apron, the extensive volume ofrough endoplasmic reticulum decreases, and allspindle cells stack within the inner retina. Thestimulus for the stacking of the spindle cells isunknown; this subclinical event indicates, however,that peripheral spindle cells are no longer thestimulus for neovascularisation and that severe ROPwill either regress or progress forunknown reasons. l1

KINETICSAt all gestational ages the moment of prematurebirth usually initiates the development of ROP. Thedevelopment may, however, be initiated beforebirth by an intrauterine catastrophy, in which casethe disease may be seen earlier than eight to10 weeks after birth. Similarly, the initiating eventmay occur weeks after premature birth if theoxidant/antioxidant balance is upset beforecompletion of the inner retinal vasoformation withdisappearance of all spindle cells, in which case thedisease may be seen eight to 10 weeks after thedelayed imbalance. Development of severe ROP isa predictable cascade of events triggered by stressedspindle cells and their subsequent synthesis andsecretion of angiogenic factors for a given length oftime.


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


Conceptional age (gestational age at birth plusweeks of life) should not be applied to the timing ofthe development of severe ROP or to the timing ofsurgical intervention. Conceptional age, however,

--2000-c

3.0- 1500,

1000

500

I

may be applied to the concept of a given posteriorarea of vascularised inner retina (ROP stage 3+)that can be dragged by an actively contractile sheetof myofibroblasts, by a fixed inactive remnant

0

0U1s

A.

n=28

1.0 1.5 2.0 2.5Horizontal diameter of globe (cm) at birth

Fig 4 Size ofeyes (horizontal diameter ofglobe (cm) at birth) is associated with birth weight (g) and appropriategestational age (weeks) at birth. The solid box indicates ocular size at birth. The dashed box indicates the predicted ocularsize 10 weeks after birth. The horizontal lines, vertical lines, and combined horizontallvertical lines are the same as in fig 3.Closed circles are preterm infants. Open circles are term infants and adults. The eyes with zone 1 ROP treated at about10 weeks have more growth yet to occur than similar eyes with zone II ROP.

10AV

10,,or

iffAV


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


i

I74E 44

E~T1h]' .l .;'2-

Fig 5 Migrating spindle cells (A and B) have adjacentplasma membranes (arrows) with no gap junctions and minimalrough endoplasmic reticulum (curved arrows). In contrast, stressed spindle cells (C and D) have adjacentplasmamembranes (arrows) with gap junctions and moderate rough endoplasmic reticulum (curved arrows). Transmissionelectron micrographs: A and C magnification of140 000, B andD magnification of32 000.

I


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


Table 1 Risk factors for retinopathy of prematurity

Clinical fact Clinical event Effect on spindle cells

Immaturity:Increased survival of infants Low birth weight, low gestational The smaller the infant, the larger the area of spindleweighing<1000 g age, surfactant cells in the inner retina with insufficient interstitial

retinol binding protein in the outer retina

Free radical damage:Acidosis Acute hypothermia, acute hypoxia, Subsequent reperfusion of retinal tissue results in

chronic hypotension (dopamnine) increased oxygen consumption

Large amounts of oxygen Better ventilation, pneumothorax, More oxygen free radicals trigger increased formationgiven bronchopulmonary dysplasia, of gap junctions between spindle cells

patent ductus arteriosus, apnoea

Increased amount of oxygen Transfusion of blood for Replacementoffetalhaemoglobinbyadulthaemoglobinreleased to tissues by adult intraventricular haemorrhage, shifts the dissociation curve so that more oxygen ishaemoglobin hyperbilirubinaemia, clinical released to the retina through the choroidal vessels

laboratory monitoring

Microbiologically Bacterial, fungal, viral systemic Ingestion of pathogens by macrophages leads to theconfirmed sepsis infections release of oxygen free radicals by macrophages

High light intensity Unusually high, continuous light in Continuous, high light damages the developingthe modern neonatal intensive photoreceptors so that they consume less of thecare unit oxygen that is arriving through the choroidal vessels

and more oxygen reaches the spindle cells

Table 2 Comparison of four randomised masked controlled clinical trials that evaluated the effect of prophylaxis withvitamin E on the suppression of development of severe ROP

Hittner et al12 Finer et al13 Johnson et al'4 Phelps et al/5

Study design:When study was conducted November 1979 to July 1978 to January 1979 to December 1980 to

December 1980 April 1981 May 1981 August 1983Year that results were published 1981 1982 1988 1987Initial route of administration Oral Intramuscular Slow intravenous Rapid intravenous

infusion, or infusionintramuscular

Form of vitamin E used Alcohol Acetate Alcohol AlcoholMean vitamin E level in study (mg%) 1-2 4-6 5.0 3 5

Infants studied (all weighed<1500 g at birth)-control/treatment groups:Mean birth weight (g) 1050/1050 1203/1185 1156/1162 1205/1181No receiving first dose of vitamin E

within 24 hours of birth 75/75 64/62 194/186 147/140No of infants whose eyes were examinedwhen they were >8 weeks of age 51/50 50/47 147/141 99/97

No of infants that died 24/24 14/15 47/45 29/31No of infants excluded 0/1 0/0 0/0 19/12Toxic effects reported None None Necrotising Intraventricular

enterocolitis, haemorrhage,sepsis retinal

haemorrhages

Outcome of ROP-control/treatment groups:ROP 2stage 3+, weighingS1500 g

at birth 5/0 3/0 7/3 1/1ROP 2stage 3+, weighing 61000 g

at birth 4/0 2/0 5/3 1/1p Value 0-01 0-01 0-05 NS


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


myofibroblast ridge, or by ocular growth notaccompanied by further retinal development, alongthe central/peripheral vector. Thus loss of retinalfunction by dragging without detachment, or byretinal detachment, is an event that is often delayedby months or years and is associated with continuoustraction for some time.

RISK FACTFORSThe risk factors of ROP are those clinical events thatstress spindle cells. These events fall into twogroups: immaturity and free radical damage(table 1)."1I Additionally, maternal angiogenicfactors caused by diabetes or hyperthyroidism mayinitiate the development of severe ROP that followsa time sequence indicating an intrauterine onset. Acombination of genetic factors may also induce aclinical appearance that mimics severe ROP butwhich is histologically distinct."l

ANTIOXIDANT: VITAMIN E SUPPLEMENTATIONThere have been four randomised, masked controlledclinical trials to evaluate the effect of prophylaxiswith vitamin E on the suppression of the develop-ment of severe ROP (table 2).12-15 The protocol foreach trial was designed so that vitamin E should begiven from the first day of life and should becontinued without interruption until inner retinalvascularisation was complete. Each used differentvitamin E preparations, achieved different plasmavitamin E concentrations, used a different initialroute of administration, evaluated ROP by differentclassification systems, set different end points foranalysis, and had different percentages of infantslost to follow up.The three clinical trials in which prophylaxis with

vitamin E was shown to be effective may beexplained by the spindle cell/myofibroblast patho-genesis of ROP. Early and continuous vitamin Esupplementation resulted in sufficient retinal uptaketo stabilise spindle cell membranes against freeradical damage in almost all infants weighing1001-1500 g at birth. This supplementation alsopartially protected most extremely high risk infants(those weighing 1000 g or less at birth) unless severalrisk factors were present. Johnson et al in theirinitial report took one year as their end point, whichallowed exclusion of a few totally blind infants in thecontrol group who died before 1 year of age.16The single trial in which prophylaxis with vitamin E

was not shown to be effective15 may also beexplained by the spindle cell/myofibroblast patho-genesis of ROP. The initial, rapid intravenousinfusion may have resulted in delayed uptake intothe retina caused by sequestration by the liver of thetransiently high plasma vitamin E concentrations.

This may have left the spindle cells in an unfavour-able oxidant/antioxidant balance for long enoughto induce formation of gap junctions. Some ofthe treated infants received doses intended forinfants in the control group in error; this interruptedthe prophylaxis. Thus even if there was initialstabilisation of the spindle cell membranes, spindlecells may have been left in an unfavourable oxidant/antioxidant balance for long enough to induceformation of gap junctions. The incidence of intra-ventricular haemorrhage was increased in thoseinfants receiving treatment. There is a close associa-tion between intraventricular haemorrhage andROP; thus the incidence of ROP in the infantsreceiving treatment may have been artificiallyincreased, which would have masked any effect ofvitamin E on the suppression of the development ofsevere ROP. This increased incidence of intra-ventricular haemorrhage is not in accordance withthe results of Speer et a117 or Sinha et al,18 and mayhave been caused by the initial rapid intravenousinfusion. Vitamin E may have displaced vitamin Kat its binding site.The issue of vitamin E supplementation to

suppress the development of severe ROP wasrecently reviewed by the Institute of Medicine of theNational Academy of Sciences.19 The committeeconcluded that there was no definite evidence ofeither benefit or harm from prophylaxis withvitamin E against ROP, and that the risks of givingvitamin E seem to be minimal for premature infantsprovided that the doses are kept low enough toachieve a blood concentration of no higher than3 mg%.Our personal clinical experience since 1979 with

vitamin E prophylaxis is more favourable. No infantwho weighed greater than 1001 g at birth, whoreceived vitamin E according to our protocol, andwho did not have severe hypoxia during delivery orsevere hypotension with subsequent reperfusion,has developed severe ROP. Thus a sufficient retinaluptake of plasma vitamin E must be stabilising mostof the spindle cell apron and halting the cascade ofevents that may ultimately result in traction andretinal detachment. The development of severeROP in infants weighing 1000 g or less at birth isdelayed. Insufficient retinal uptake of plasmavitamin E caused by there being insufficient amountsof interstitial retinol binding protein in the subretinalspace may only transiently suppress formation ofgap junctions between spindle cells in the innerretina. Thus in the smallest and illest infants,prophylaxis with vitamin E may ultimately fail toprevent sufficient spindle cell membrane damage.Severe ROP develops about two weeks later than insimilar infants not receiving vitamin E.


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/



on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


The overall incidence of all stages of ROP is notdecreasing because the protection given to plasmamembranes by vitamin E is not an 'all or nothing'effect, and is by no means a panacea. Vitamin Eprophylaxis is drastically decreasing the severity ofROP in infants weighing between 1001 and 1500 g atbirth.The ultimate clinical question is whether vitamin E

prophylaxis to decrease the development of severeROP can be justified for all infants weighing 1500 gor less at birth. The issue of its safety has been at theroot of the controversy. With our protocol we havenot seen a single case of toxicity from vitamin Eprophylaxis in our centre since we began giving it in1979. All reported toxic reactions20 have resultedfrom high peaks of plasma vitamin E given by rapidintravenous administration (up to 15 mg%), or high,sustained plasma vitamin E concentrations achievedby slow intravenous infusion or intramuscularinjection (more than 8 mg% for several days), or thehigh osmolality of oral vitamin E preparations(more than 2000 mOsm), or the use of impropercarriers in small amounts to make intramuscularpreparations soluble (sesame seed oil, EmulphorE1-100, and polysorbate 80), or the use of toxicpolysorbate carriers in large amounts in E-ferol.E-ferol is a drug that was illegally sold in the UnitedStates for a short time from late 1984 to early 1985;it was recalled after many deaths had been reported,for which civil and criminal actions were taken.

OUR CURRENT RECOMMENDATIONS CONCERNINGVITAMIN ETo suppress the development of severe ROP, wegive 100 mg/kg/day of dl-a-tocopherol (or dl-a-tocopheryl acetate) orally in a medium chain trigly-ceride solution from the first hours of life untilretinal vascularisation is complete. To suppress thedevelopment of severe intraventricular haemor-rhage,17 18 we give 15 mg/kg intramuscularly or5 mg/kg intravenously over eight hours as soon aspossible after delivery, and 10 mg/kg intramuscularlyor 3 mg/kg intravenously over eight hours on thesecond day of life. To maintain mean plasma

vitamin E concentrations between 1-2 and 3-5 mg%until oral feedings can be started (or if oral feedingsare interrupted) we give 10 mg/kg intramuscularlyevery third day or 3 mg/kg/day intravenously overeight hours. We use only dl-a-tocopherol in water.

ANTIOXIDANT: SELENIUM SUPPLEMENTATIONBecause vitamin E provides incomplete, agedependent, antioxidant protection against thedevelopment of severe ROP, future research mustfocus on antioxidant systems that are alreadyendogenous to the retina of the infant weighing1000 g or less at birth and that can be stimulated bywater soluble substances. Vitamin C is an antioxidantsystem that meets these criteria; the system is,however, already operating maximally in infants ofall gestational ages.6 Two independent laboratorieshave documented high retinal concentrations ofselenium dependent glutathione peroxidase in allpremature infants weighing 160 g or more atbirth.21 22 Perhaps selenium supplementation fromthe first hours of life will maintain the activity of thisantioxidant system and protect spindle cells in allinfants who are likely to survive.

Traction (myofibroblasts)

REGRESSION OF STAGES 1, 2, OR MILD STAGE 3 ROPWhen regression of mild ROP (international classi-fication of ROP stages 1, 2, or mild stage 3) occurs,synthesis and secretion of angiogenic factors stop,gap junctions between adjacent spindle cellsdisappear uniformly throughout the apron, theextensive volume of rough endoplasmic reticulumdecreases, and spindle cells remain as an apron.When regression of mild ROP occurs, inner retinalvasoformation continues by canalisation and reachesthe ora serrata. Peripheral and posterior retinal andvascular changes tabulated by international classi-fication of ROP10 do not follow stages 1 or 2 ROP,and there is no evidence of any remnants ofmyofibroblasts. These changes are minimal aftermild stage 3 ROP, and there are minimal remnantsof myofibroblasts, but usually no appreciable

Fig 6 Top: preoperative appearance at the age of10 weeks ofright retina ofan infant weighing 650 g at birth with zone IIstage 3+ ROP. Note the extent ofvascularised retina (distance between the pair ofthree vertical dashes equals 2*5 cm ofretina); and immature macula (circle); and the distance between the temporal rim ofthe optic disc (two vertical dashes) andthe centre ofthe macula equals 0-5 cm ofretina). Moderate plus disease and moderate invasion ofmyofibroblast are present.Bottom: postoperative appearance ofthe same eye at the age of10 months. Note extent of viable retina (distance between thepair ofthree double vertical dashes equals 1 9 cm ofretina); temporally displaced macula (circle); and distance between thetemporal rim ofthe optic disc (two vertical dashes) and centre ofthe macula equals 0-7 cm ofretina. Dragging ofposteriorvessels and the region offull thickness cryotherapy are visible. Fixation is unsteady (nystagmus) in this myopic (-9 00) eyeat the age of21 months. Comparison ofthese two retinal montages shows that cryotherapy to the avascular retina and shuntalone do not guarantee afavourable outcome with respect to macularfunction.


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/



on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


dragging of the larger retinal vessels, or of themacula.

REGRESSION OF SEVERE STAGE 3 ROPWhen regression of severe ROP (internationalclassification of ROP moderate or severe stage 3)occurs, synthesis and secretion of angiogenic factorsstop, gap junctions between adjacent spindle cellsdisappear uniformly throughout the apron, theextensive volume of rough endoplasmic reticulumdecreases, and all spindle cells stack within the innerretina. The stimulus for the spindle cell stacking isunknown; this subclinical event indicates, however,that peripheral spindle cells are no longer thestimulus for neovascularisation and that severe ROPwill either regress or progress for unknown reasons.When regression of severe ROP occurs, innerretinal vasoformation may occur between the stackedspindle cells, but this vasoformation is not bycanalisation and does not reach the ora serrata. Insome cases of regression of severe ROP, greyishremnants of the contractile myofibroblast sheetsthat are at the site of the former shunt are leftbehind along the retina vitreous surface as the onlysequelae of the active ROP tabulated by internationalclassification ofROP as mild peripheral, and posteriorretinal and vascular changes.10 In other cases thesegreyish remnants exist in sufficient numbers to causedragging of the larger retinal vessels and of themacula. This is usually seen clinically as temporaldragging of the vessels and macula, tabulated byinternational classification of ROP as moderateperipheral, and posterior retinal and vascularchanges. 10

PROGRESSIONWhen progression of severe ROP occurs, synthesisand secretion of angiogenic factors stop, and allspindle cells stack. Clinically the posterior retinalvessels become dilated and tortuous (which istermed 'plus disease' according to internationalclassification of ROP9), the shunt becomes engorged,and the myofibroblast sheets become vascularisedand eventually haemorrhage into the vitreous (fig 6

top, fig 7 top, fig 8 top). Histologically a surge ofmyofibroblast sheets differentiate from the shuntand invade the vitreous (fig 9).

THE CONCEPr OF CRYOTHERAPYCryotherapy for ROP entails the application of adestructive cold probe to the scleral surface. Thecold application must destroy both the spindle cellsin the inner retinal layer and the myofibroblasts onthe retina/vitreous surface. Cryotherapy destroys allliving ocular tissue between the sclera and the retina/vitreous surface round the complete circumferenceof the eye. Fibroblasts from adjacent living sclerainvade the dense scleral connective tissue residue,and allow anterior enlargement of the globe aftercryotherapy.

CRYOTHERAPY TO THE AVASCULAR RETINA FORSTAGE 2 OR MILD STAGE 3 ROPCryotherapy to the avascular retina when onlyneovascularisation is present, (that is when spindlecells are still secreting angiogenic factors andmyofibroblasts have not separated from the shunt)has been suggested. Theoretically, obliteration ofthe peripheral inducers would stop the progressionof the disease completely. Such early cryotherapy,however, may cause problems in the smallest infants.Many sessions of cryotherapy would be needed toobliterate spindle cells that were intermingled withthe formed vessels at the time of the initialcryotherapy but which continue to migrate into thetreated areas later, and there would be destructionof an appreciable amount of the peripheral visualfield and possibly macular dragging in an eye thatmight have undergone spontaneous regression.Thus early cryotherapy (at ROP stage 2 or mildstage 3) is not recommended.

CRYOTHERAPY TO THE AVASCULAR RETINA FORMODERATE OR SEVERE STAGE 3 ROPThe National Eye Institute of the National Institutesof Health in the United States of America sponsoreda clinical trial to evaluate the effect of cryotherapyto the avascular retina for ROP in eyes randomised

Fig 7 Top: preoperative appearance at the age of10 weeks ofthe right retina ofan infant weighing 880 g at birth, with zoneII stage 3+ ROP. Note the extent of vascularised retina (distance between the pair ofthree vertical dashes equals 2-4 cm ofretina); and immature macula (circle); and the distance between the temporal rim ofthe optic disc (two vertical dashes) andthe centre ofthe macula equals 0-5 cm of retina. Severe plus disease and an area ofhaemorrhage adjacent to severe invasionofmyofibroblasts are visible. Bottom: postoperative appearance ofthe same eye at the age of20 months. Note temporallydisplaced, abnormally pigmented macula (circle), the distance between the temporal rim ofthe optic disc (two verticaldashes) and the centre ofthe macula equals 0-6 cm ofretina, and the anterior extent ofthe therapeutic scleral buckle (widesolid circular line). Fixation is unsteady (nystagmus) in this myopic (-9.50) eye. Comparison ofthese two retinal montagesshows that cryotherapy to the avascular retina and shunt alone did notprevent retinal detachment, and that a therapeuticscleral buckle does notpermit afavourable outcome with respect to macularfunction.


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/



on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


Fig 9 Transmission electron micrograph ofa myofibroblast with actin filaments (arrows) criss crossing the cytoplasm andending in peripheral densities (curved arrows) that are analogous to Z lines ofstriated muscle (magnification of48 000).

in symmetrical disease or in randomised infants inasymmetrical disease. Threshold disease was thedevelopment of five or more contiguous or eightcumulative clock hours of stage 3 ROP in zone I orII in the presence of 'plus' disease. This clinical trialfound that cryotherapy to the avascular retina (asdetermined by fundus photographs of outcomethree months after treatment) reduced the numberof posterior retinal detachments, retinal folds affect-ing the macula or retrolental tissue, by 50%.23 Of156 treated eyes, 22% had an unfavourable outcome(78% had a favourable outcome), and of 149untreated eyes, 43% had an unfavourable outcome(57% had a favourable outcome). Thus neithervitamin E nor cryotherapy is a panacea.Though this collaborative trial evaluated the

effect of cryotherapy for ROP, many questions stillremain unanswered. The mean birth weight for thestudy was 801 g and only 12 eyes (8%) had zone IROP. It is unfortunate that the study will generatelong term data only for comparatively large infants(92% had zone II ROP). According to the spindlecell/myofibroblast pathogenesis of ROP, if vitamin Esupplementation for prevention of severe ROP hadbeen given, many of the larger preterm infantsmight not have required cryotherapy, and the ratioof zone I: zone II disease would have been moreequal. With increasing survival of the smallestpreterm infants, however, the dilemmas of zone Idisease cannot be addressed by the data base of thistrial. Unfortunately the outcomes of zone I, whichare more difficult to treat, were grouped with those

Fig 8 Top: preoperative appearance at the age of10 weeks ofthe right retina ofan infant weighing 760 g at birth with zoneII stage 3+ ROP. Note the extent of vascularised retina (distance between the pair of three vertical dashes equals 2-4 cm ofretina); and immature macula (circle); and the distance between the temporal rim ofthe optic disc (two vertical dashes) andthe centre ofthe macula equals 0-5 cm ofretina. Severe plus disease and an area ofhaemorrhage adjacent to severe invasionofmyofibroblasts are visible. Bottom: postoperative appearance ofthe same eye at the age of17 months. Note undisplaced,normally pigmented macula (circle), the distance between the temporal rim ofthe optic disc (two vertical dashes) and thecentre ofthe macula equals 0-5 cm ofretina, and the anterior extent ofthe prophylactic scleral buckle (wide dashed circularline). Fixation is steady (no nystagmus) in this myopic (-6.25) eye at the age of18 months. Comparison ofthese two retinalmontages shows that cryotherapy to the avascular retina and shunt combined with a prophylactic scleral buckle does permitafavourable outcome with respect to macularfunction.


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


of zone II. The study reported treatment that rangedfrom 6-7 to 23-9 weeks. The timing of cryotherapywas at a mean of 114 weeks, however, whichcoincides with the window of treatment (generally8 to 10 weeks, but occasionally 6 to 15 weeks)predicted by the spindle cell/myofibroblast patho-genesis of ROP. At the later time points (15 weeksor more), it is not stated how many eyes weretreated or what the percentage of favourable out-comes was. Unfortunately the study grouped thepercentage of favourable outcomes without regardto the timing of cryotherapy. Thus it is unknownhow many eyes were treated primarily to preventtraction associated with myofibroblasts. Despite aprotocol designed to avoid the shunt, appreciablehaemorrhages occurred in 30 eyes (19%). This highincidence was most likely caused by accidentalcryotherapy to the shunt while it was still engorgedwith blood. Though it was important to expeditepublication of the effectiveness of cryotherapy, only45 infants had been followed for one year. Becauseof the developmental process of ocular growth(fig 4), the spindle cell/myofibroblast pathogenesisof ROP would predict that many of the reportedsuccesses will become failures with long term followup. Tasman's editorial about the study has alreadyalluded to this problem by describing rhegmato-genous retinal detachment in the better eye of twopatients (at 1 and 4 years of age).24 He attributesthese retinal detachments to the intolerable stressassociated with ocular growth that occurs aftercryotherapy caused by a firm peripheral chorio-retinal adhesion in small eyes. Our centre had asimilar retinal detachment in the better eye of apatient (at 6 years of age).25We realise that when the remnant myofibroblast

sheets produce minimal traction, the single procedureas recommended by the collaborative group may besufficient; the larger retinal vessels and macula can,however, be appreciably dragged by ocular growth,and visual acuity may be as low as 20/200. Frequently,when myofibroblast sheets produce appreciabletraction, this single procedure is not sufficient tocounterbalance the forces generated by an activecontraction of myofibroblast sheets, by a fixedremnant myofibroblast ridge, or by ocular growth,and retinal detachment can occur. To obtain retinalreattachment, a therapeutic scleral buckle is required(often with drainage of subretinal fluid). Scleralbuckling entails placing a thin (0.75 mm) but broad(3-5 mm) silicone band round the equator of theeye under the four rectus muscles. Even when theretina is successfully reattached anatomically, themacular region has a distorted clinical appearance,and the best visual acuity that can be anticipated is20/400.

OUR CURRENT SURGICAL PROTOCOLWith our long experience of cryotherapy to theavascular retina (since 1974) and with the unfavour-able visual outcome of most of the patients causedby macular dragging or retinal detachment, we haveadvocated since 1987 the following surgical opera-tions based on both the spindle cell/myofibroblastpathogenesis ofROP and the reality ofocular growth.

Destroy the spindle cell apron that is the source ofangiogenic factors. This part of our current protocolis identical to that recommended by the collaborativestudy except thatwe treat both eyeswhen symmetricaldisease is present. Clinically, within a day or twoafter cryotherapy to the avascular retina, the posteriorretinal vessels become less dilated and tortuous, theshunt empties of blood, and the sheets of myofibro-blasts atrophy.

Destroy the shunt and sheets of myofibroblaststhat are the source of traction leading to maculardragging or retinal detachment. This second cryo-therapy to the shunt and sheets of myofibroblasts(optimally some three to seven days after the initialcryotherapy to the avascular retina) will produceless haemorrhage than might be produced if theshunt and sheets of myofibroblasts are inadvertentlytreated during the initial cryotherapy. The twotreatments may be sufficient to prevent retinaldetachment in some cases, but do not preventdragging of the macula (fig 6 bottom). The twocryotherapies, however, may not in many cases besufficient to prevent retinal detachment, and atherapeutic scleral buckle may be required toreattach the retina (fig 7 bottom).

Place a prophylactic scleral buckle at the site ofthe former shunt at the time of the second cryo-therapy. With continued growth of the anteriorportion of the eye, this band changes the generalshape of the eye from a sphere to an ellipse so thatthe posterior portion of the eye in which the macularesides is not dragged. Additionally, this bandcounterbalances tractional forces from the activecontraction of sheets of myofibroblasts or from afixed inactive remnant myofibroblast ridge that canlead to retinal detachment even when the shunt hasbeen obliterated.

This staged procedure seems to allow the maculato develop maximally without distortion fromdragging or detachment, and visual acuity is usuallybetter than 20/40 (fig 8 bottom). When myofibro-blasts have completely disappeared from the vitreous,complete quiescence of ROP has been obtained.When ocular growth is complete (at about18 months), it is usually safe to remove the scleralbuckle and permit the eye to return to its normalspherical shape.As successful as these drastic operations may be,


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/


this also is not a panacea, and is not the ultimatesolution to the continuing saga of ROP. Thus thechallenge remains to stabilise spindle cells beforethe initiation of the cascade of events recountedhere-the spindle celllmyofibroblast pathogenesis ofROP as modified by ocular growth.

For the last 10 years this research has been supported by grantsfrom the Retina Research Foundation, National Institutes ofHealth (EY-02607), Hoffmann-LaRoche Inc, Research to PreventBlindness, The March of Dimes Birth Defects Foundation, TheRetinitis Pigmentosa Foundation Fighting Blindness, The FarishFoundation, The Cullen Foundation, and The Lion's Eyes of TexasEye Bank. The light and transmission electron micrographs weretaken by Rekha Mehta and Evelyn Brown. We acknowledge thephotographic expertise of Alex Kogan and Gilma Miranda.

References

Hittner HM, Rudolph AJ, Kretzer FL. Suppression of severeretinopathy of prematurity with vitamin E supplementation:ultrastructural mechanism of clinical efficacy. Ophthalmology1984;91:1512-23.

2 Kretzer FL, Mehta RS, Johnson AT, Hunter DG, Brown ES,Hittner HM. Vitamin E protects against retinopathy ofprematurity through action on spindle cells. Nature 1984;309:793-5.Kretzer FL, McPherson AR, Rudolph AJ, Hittner HM. Patho-genic mechanism of retinopathy of prematurity: a controversialexplanation for the efficacy of oral and intramuscular vitamin Esupplementation and cryotherapy. Bull NY Acad Med 1985;61:883-900.

4 Kretzer FL, McPherson AR, Hittner HM. An interpretation ofretinopathy of prematurity in terms of spindle cells: relationshipto vitamin E prophylaxis and cryotherapy. Graefe's Archive forClinical and Experimental Ophthalmology 1986;244:205-14.

5 Johnson AT, Kretzer FL, Hittner HM, Glazebrook PA, BridgesCDB, Lam DMK. Development of the subretinal space in thepreterm human eye: ultrastructural and immunocytochemicalstudies. J Comp Neurol 1985;233:497-505.

6 Nielson JC, Naash MI, Anderson RE. The regional distributionof vitamins E and C in mature and premature human retinas.Invest Ophthalmol Vis Sci 1988;29:22-6.Kretzer FL, Hittner HM. Spindle cells and retinopathy ofprematurity: interpretations and predictions. In: Flynn JT,Phelps DL, eds. Birth defects: original article series. Retinopathyof prematurity: problem and challenge. 1988;24:147-68.

8 Hittner HM, McGee JK, Kretzer FL. The relationship betweenbirth weight which is appropriate for gestational age and thehorizontal extent of inner retinal vasculature in the very preterminfant. Invest Ophthalmol Vis Sci 1988;29:381.The Committee for the Classification of Retinopathy ofPrematurity. An international classification of retinopathy ofprematurity. Arch Ophthalmol 1984;102:1130-4.

'0 The International Committee for the Classification of the LateStages of Retinopathy of Prematurity. An international classi-

fication of retinopathy of prematurity: It. The classification ofretinal detachment. Arch Ophthalmol 1987;105:906-12.Hittner HM, Kretzer FL. Differential diagnosis of retinopathyof prematurity. In: McPherson AR, Hittner HM, Kretzer FL,eds. Retinopathy of prematurity: current concepts and contro-versies. Toronto: BC Decker, 1986:53-65.

12 Hittner HM, Godio LB, Rudolph AJ, et al. Retrolentalfibroplasia: efficacy of vitamin E in a double-blind clinical studyof preterm infants. N Engl J Med 1981;305:1365-71.

13 Finer NN, Grant G, Schindler RF, Hill GB, Peters KL. Effectof intramuscular vitamin E on frequency and severity ofretrolental fibroplasia: a controlled trial. Lancet 1982;i:1087-91.

14 Johnson L, Quinn GE, Abbasi S, Otis C, Bowen FW. VitaminE and retinopathy of prematurity. Pediatrics 1988;81:329-31.

15 Phelps DL, Rosenbaum A, Isenberg SJ, Leake RD, Dorey FJ.Tocopherol efficacy and safety for preventing retinopathy ofprematurity: a randomized, controlled, double-masked trial.Pediatrics 1987;79:489-500.

16 Schaffer DB, Johnson L, Quinn GE, Weston M, Bowen FW.Vitamin E and retinopathy of prematurity: follow-up at oneyear. Ophthalmology 1985;92:1005-11.

7 Speer ME, Blifeld C, Rudolph AJ, Chadda P, Holbein MEB,Hittner HM. Intraventricular hemorrhage and vitamin E in thevery low birth weight infant: evidence for efficacy of earlyintramuscular administration. Pediatrics 1984;74:1107-12.

18 Sinha S, Davies J, Toner N, Bogle S, Chiswick M. Vitamin Esupplementation reduces frequency of periventricular haemorr-hage in very preterm babies. Lancet 1987;i:466-71.

19 Committee of the Institute of Medicine. Report of a study-vitamin E and retinopathy ofprematurity. IOM-86-02. Washing-ton, DC: National Academy Press, 1986.

20 Hittner HM, Kretzer FL. Toxicity of vitamin E in preterminfants. In: McPherson AR, Hittner HM, Kretzer FL, eds.Retinopathy of prematurity: current concepts and controversies.Toronto: BC Decker, 1986;111-6.

21 Naash MI, Nielsen JC, Anderson RE. Regional distribution ofglutathione peroxidase and glutathione-s-transferase in adultand premature human retinas. Invest Ophthalmol Vis Sci1988;29:149-52.

22 Hittner HM, Kretzer FL, Lane HW. Selenium dependentglutathione peroxidase in preterm human retinas. Fed Proc1987;46:566.

23 Cryotherapy for Retinopathy of Prematurity CooperativeGroup. Multicenter trial of cryotherapy for retinopathy ofprematurity-preliminary results. Arch Ophthalmol 1988;106:471-9.

24 Tasman W. Multicenter trial of cryotherapy for retinopathy ofprematurity-editorial. Arch Ophthalmol 1988;106:463-4.

25 McPherson AR, Hittner HM, Kretzer FL. Treatment of acuteretinopathy of prematurity with cryotherapy. In: McPhersonAR, Hittner HM, Kretzer FL, eds. Retinopathy ofprematurity:current concepts and controversies. Toronto: BC Decker, 1986:161-78.

Correspondence and requests for reprints to Dr Frank L Kretzer,Cullen Eye Institute, Baylor College of Medicine, One BaylorPlaza, Houston, Texas 77030, USA.

Accepted 29 June 1988


on Septem


j.com/

Arch D


pec_No.1151 on 1 O

ctober 1988. Dow

nloaded from

http://adc.bmj.com/

Documents

Retinopathy - Archives of Disease in Childhood · Archives ofDisease in Childhood, 1988, 63, 1151-1167 Personal paper Retinopathy of prematurity: clinical implications of retinal