Ro

JCa

b

a

ARRAA

KRRRFMF

1

oiNittTai

ttE

bw

Af

0d

Reproductive Toxicology 26 (2008) 262–266

Contents lists available at ScienceDirect

Reproductive Toxicology

journa l homepage: www.e lsev ier .com/ locate / reprotox

etinal folding in the term rabbit fetus—Developmental abnormalityr fixation artifact?

ulian Frencha,∗, Jane Hallidayb, Marietta Scottb, Debbie Adkinsa,arsten Liessb, John C. Watertonb, Jane Stewarta

Department of Global Safety Assessment, AstraZeneca, Alderley Park, Macclesfield, UKDepartment of Imaging, Translational Sciences, AstraZeneca, Alderley Park, Macclesfield, UK

r t i c l e i n f o

rticle history:eceived 21 June 2008eceived in revised form 14 August 2008ccepted 21 August 2008vailable online 28 August 2008

eywords:eproductive toxicology

a b s t r a c t

Preclinical reproductive toxicology studies must be performed to provide information on the risk of caus-ing fetal harm in pregnant patients. These studies detect fetal malformations, which may or may not bedrug-related adverse events. In the rabbit fetus, “slight retinal folding” is commonly observed; there isanecdotal evidence that these folds may be caused by routine (Bouin’s fluid) fixation. This study usedmagnetic resonance imaging (MRI) to assess rabbit retinal architecture in fresh specimens, which wasthen reassessed following Bouin’s fluid fixation. A total of 30 fetuses from 5 litters were imaged. No reti-nal folding was detected in fresh specimens but it was observed in a majority of fetuses post-fixation. Theuse of Davidson’s fixative followed by Bouin’s fluid showed a markedly lower incidence of “slight retinal

etinal folding

abbitetal

folding”. Conclusion: slight retinal folds in the rabbit fetus are likely artifactual and can be reduced usingDavidson’s fixation prior to Bouin’s.

itmacdmpt

dopia

RIixatives

. Introduction

Studies are performed in pregnant animals to assess the effectsf test compounds on embryofetal development and hence providenformation on risk of causing fetal harm in pregnant patients. Theew Zealand white rabbit is routinely used in studies of this kind,

n which pregnant rabbits are dosed with candidate drugs duringhe organogenesis phase of gestation. Fetuses are removed nearhe end of the pregnancy and examined for developmental defects.hese defects include minor fetal malformations whose physicalppearance may be a subtle deviation from normal. One such defects “retinal folds”.

Fig. 1 depicts the anatomy of the eyeball. Folding or buckling ofhe retina can result in various degrees of vision loss. In the humanhis is caused by conditions such as Norrie’s disease [1] or Familial

xudative Vitreoretinopathy [2].

Retinal folds have been recorded in near term rabbit fetuses inoth slight and severe forms (Fig. 2), the severe form being rare,ith a spontaneous fetal incidence of 0.9%; however, the slight form

∗ Corresponding author at: AstraZeneca Pharmaceuticals, 23F89A Mereside,lderley Park, Macclesfield, Cheshire SK10 4TG, UK. Tel.: +44 1625 519163;

ax: +44 1625 516809.E-mail address: julian.french@astrazeneca.com (J. French).

tha

ncDbbs

890-6238/$ – see front matter © 2008 Elsevier Inc. All rights reserved.oi:10.1016/j.reprotox.2008.08.008

© 2008 Elsevier Inc. All rights reserved.

s much more common and has been recorded at incidences of upo 45%, in fixed samples, in this laboratory (unpublished data, litter

ean incidence recorded in control group specimens (n = 80) fromrecent study). The high incidence of slight retinal folds observed inontrol group fetuses, as well as those exposed to test compounduring organogenesis, suggests this is either a common develop-ental variation in the rabbit or that it may be an artifact of the

rocedures used in the handling, fixing or observing of the fetalissue.

Currently, many laboratories use a protocol in which fetuses areecapitated and the heads fixed in Bouin’s fluid for a minimumf 2 weeks. Bouin’s fluid penetrates tissues well, fixes rapidly, androvides a permanent yellow colour to the fixed tissue so provid-

ng clarity in distinction between structures. It is also sufficientlycidic to act as a decalcifying agent, which facilitates sectioninghrough skull bones. Coronal sections are prepared using a free-and blade technique, and the eye lens and vitreous removed tollow the retinal surface to be examined.

Although the use of Bouin’s fluid has many advantages, it mayot be the most appropriate fixative to use in the assessment of

ranial structures. Alternatives, using either buffered formalin oravidson’s fixative, alone or in combination with Bouin’s fluid, maye more suitable. Artifactual, cyst-like spaces present in fetal rabbitrain tissue, observed following fixation in Bouin’s fluid, were noteen if the heads were initially fixed in formalin then changed into

J. French et al. / Reproductive Toxicology 26 (2008) 262–266 263

Bsmseaac

fctfnead

trop

2

socrpftes

paowua

effwwlr

Fff

flwo

5a4

After MRI examination, the heads were fixed, at room temperature, for a min-

Fig. 1. The anatomy of the eyeball in Bouin’s fixed coronal section and MRI.

ouin’s fluid [3]. As this fixation regime was apparently respon-ible for reducing the incidence of an artifactual observation itay be a better method than Bouin’s alone. Alternatively, David-

on’s fixative (which is used routinely for optimum fixation ofyes from adult animals [4] in general toxicology studies), eitherlone or for 2 weeks, followed by a week in Bouin’s fluid, could benother fixation method which is less prone to inducing artifactualhanges.

It has already been suggested that slight retinal folds are arti-actual in origin and may form during the period of fixation; theurrent draft version of the IFTS (International Federation of Tera-ology Societies) international glossary of terms notes that “retinalolds may be due to processing artifact” [5], however, there iso experimental proof of this. Obviously, establishing the exacttiology of this finding is important in the risk assessment forny candidate drug, to enable the correct interpretation of theata.

This study uses MRI in conjunction with freehand blade dissec-ion and aims to provide definitive evidence as to whether the slightetinal folds observed in Bouin’s fluid fixed rabbit fetuses are devel-pmental or artifactual in origin. Supplementary studies were alsoerformed to investigate the use of alternative fixatives.

. Materials and methods

Sexually mature, virgin female New Zealand white rabbits (substrain HsdifNZW,upplied by Harlan UK Ltd., Belton, Leicestershire) were paired with unrelated malesf the same strain. Animals were maintained under standard husbandry/enrichmentonditions and provided with drinking water and pelleted food (Teklad 2030abbit diet) ad libitum, and irradiated hay. The study was performed in full com-liance with licences issued under the Animal (Scientific Procedures) Act 1986ollowing local ethical committee review. Power calculations were performedo determine the minimum number of fetuses and litters required for the MRIxaminations, based on seeing a 20% change between pre- and post-fixation mea-urements.

Animals were killed on Day 29 post-coitum by intravenous injection of sodiumentobarbital, and a post-mortem examination performed. The uterus was removednd any live fetuses present were removed, weighed, and killed using an injectionf sodium pentobarbital. Fetuses were examined externally and decapitated. Headsere identified by uterine position and then kept on wet ice until ready for scanningsing MRI. A total of 30 fetuses from 5 litters were examined, and MR images werecquired both pre- and post-fixation in Bouin’s fluid.

All imaging was performed on a Varian Inova 9.4 T horizontal bore scannerquipped with 400 mT/m gradients and a 25 mm diameter transmit/receive sur-ace coil. The specimen was placed at the centre of the magnet bore, with the eyes

acing towards the back of the magnet. The coil was positioned so that it sat, levelith the eyes, like a halo around the rabbit’s head. A fiducial tube containing wateras positioned next to the right eye. Manual shimming was performed so that water

ine widths of 60 ± 10 and 68 ± 14 Hz were achieved for the fresh and fixed samples,espectively. Multi-slice 2D gradient-echo scout images (TR = 200 ms, TE = 5.9 ms,

i5aut

ig. 2. Comparison of a normal retina (A) with those showing slight (B) or severe (C)olding from Bouin’s fluid fixed fetuses collected in a previous study (unpublished)rom this laboratory. Note that the lens has been removed in specimens (A) and (B).

ip angle = 30◦ , 2 ms sinc pulse for excitation, 128 × 128 matrix, 10 slices, 1 average)ere acquired in all three orientations to ensure accurate positioning of the volume

f interest for 3D image acquisition.T1-weighted 3D gradient-echo images (TR = 13.9 ms, TE = 5.9 ms, flip angle = 40◦ ,

0 �s hard pulse for excitation, 128 × 128 × 64 matrix, 24 averages) of the eyes werecquired with an isotropic resolution of 210 �m. The total measurement time was5 min.

mum of 2 weeks in Bouin’s fluid (15 parts picric acid saturated aqueous solution,parts 40% formaldehyde solution A.R. v/v, 1 part glacial acetic acid). Post-fixation,further MR image of each head was acquired using identical parameters to thosesed pre-fixation with the exception of the flip angle, which was set to 60◦ in ordero obtain optimum tissue contrast. Post-acquisition, images were reconstructed and

2 e Toxicology 26 (2008) 262–266

sf

tw

oetfsae‘b

fioe

efia

2

3

4

t

3

appstnf

caffs

ihtf“tNfii

iiflett

F(f

hfnfid

64 J. French et al. / Reproductiv

aved in Analyze format (Mayo Clinic, USA) and examined for the presence of retinalolds using ImageJ (National Institutes of Health, USA).

Coronal sections were prepared from the fixed heads using a freehand bladeechnique [6]. The lens and vitreous humour were removed and the retinal surfaceas examined using a low power (8×) stereomicroscope.

Corresponding MRI scans of the fresh and fixed heads were compared with eachther, as well as with the head sections, for the occurrence of retinal folds in eitherye. The specimens were analyzed independently by three different individuals forhe presence of retinal folds—one looking at the MRI data and two looking at thereehand blade sections. This was to reduce any bias that may result due to theubjectivity in counting folds. The differences between the three analyses have beenssessed using McNemar’s change test [7], which is appropriate for situations whereach subject is used as its own control, typically with a measurement ‘before’ and

after’ some treatment. The three methods are considered in pairs and the changeetween each pair of methods for each fetus is analyzed.

The difference between the incidence before fixation and the incidence afterxation was also assessed using McNemar’s change test. The presence or absencef folding in each fetus after fixation was compared to the pre-fixing incidence. Inach case, McNemar’s test was performed at the 5% level.

To supplement this work, further investigations have examined the use of differ-nt fixation processes. Fetal heads (18–54 heads per technique, detailed below) werexed using one of four methods and coronal sections were prepared and examineds described above:

1. 54 heads fixed using: Bouin’s fluid alone for 2 weeks.. 18 heads fixed using: 10% buffered formalin v/v (1 part formalin, 9 parts dis-

tilled water, pH range 6.8–7.2, buffered using sodium dihydrogen orthophosphate(4 g/l) and disodium hydrogen orthophosphate (6.5 g/l)) for 2 weeks followed byBouin’s fluid for 1 week.

. 33 heads fixed using: Davidson’s fixative (3 parts methylated spirits 740P, 2 parts40% formaldehyde solution v/v, 1 part acetic acid, 3 parts distilled water) for 2weeks followed by Bouin’s fluid for 1 week.

. 20 heads fixed using: Davidson’s fixative alone for 2 weeks.

The incidence of slight retinal folding for each fixation regime was recorded andabulated.

. Results

Representative slices from the 3D MR images of the freshnd fixed fetuses can be seen in Fig. 3A and B, respectively. Ahotograph of the dissected eye is shown in Fig. 3C. All fetusesrovided analyzable MR images. Using the pulse sequences cho-en, the MR signal intensities are in the order vitreous > periorbitalissues > retina ≥ lens. Fig. 3B and C shows that the MRI tech-ique used had sufficient resolution to allow detection of retinal

olds.Table 1 shows the incidence of retinal folding from stereomi-

roscope observation of the freehand blade section (observers 2nd 3) and the MR images (observer 1) in the Bouin’s fluid fixedetuses. Table 2 shows the incidence of retinal folding for theour different fixation methods, as determined by freehand bladeection.

All the folds reported were classed as “slight” and the tablesnclude all folds detected. However, folds occurring at the foveaave been recorded separately from those occurring elsewhere onhe retina as, according to current guidelines in this laboratory,oveal folds would be excluded from the general observation ofslight retinal folds”. Therefore, the statistical analysis includes onlyhose folds occurring away from the foveal region of the retina.o retinal folding could be detected using MRI in any of the fresh

etuses, with folds observed in 78 ± 2% of samples post-fixationn Bouin’s fluid alone (p < 0.001 as compared to the pre-fixationncidence of zero).

High incidences of both retinal and foveal folds were recordedn fetuses fixed in Bouin’s fluid alone. The heads of fetuses fixed

nitially in formalin for 2 weeks, followed by 1 week in Bouin’suid contained crystalline deposits which made sectioning andxamination very difficult. For the small number of these headshat were sectioned there was no apparent difference betweenhe incidences of folds observed, in either Bouin’s fluid alone fixed

btfes

ig. 3. Representative slices from 3D MR images of fresh (A) and Bouin’s fluid fixedB) eyes plus photographic image of the coronally sectioned eye (C) from the sameetus in this study. Note that the lens has been removed in specimen (C).

eads or formalin/Bouin’s fluid fixed heads. Further sectioning oformalin/Bouin’s fluid fixed heads was, therefore, abandoned. Aso appreciable decalcification was apparent in Davidson’s fixativexed heads, sectioning was impossible and this too was aban-oned.

Fetuses fixed initially in Davidson’s fixative for 2 weeks, followed

y 1 week in Bouin’s fluid could be sectioned without difficulty andhe appearance of the sections was superficially indistinguishablerom those prepared from a fetus fixed in Bouin’s fluid alone. How-ver, the incidence of foveal folds was reduced to a tenth of thateen with Bouin’s fluid alone while the incidence of retinal folds in

J. French et al. / Reproductive Toxicology 26 (2008) 262–266 265

Table 1Retinal folds in fixed fetal specimens: litter incidence and retinal location

Animal identification Number (%) of fetuses displaying folds in the foveal region Number (%) of fetuses displaying retinal folds in other areas

Observer 1,MR Image

Observer 2,coronal section

Observer 3,coronal Section

Observer 1,MR image

Observer 2,coronal section

Observer 3,coronal section

Dam a 3/4 4/4 4/4 3/4 3/4 3/4Dam b 6/6 5/6 5/6 5/6 6/6 6/6Dam c 7/10 7/10 6/10 5/10 6/10 4/10Dam d 4/4 4/4 4/4 4/4 4/4 4/4Dam e 3/6 3/6 2/6 5/6 4/6 4/6

Total 23/30 (79%) 23/30 (81%) 21/30 (75%) 22/30 (78%) 23/30 (80%) 21/30 (76%)

Average % 78 78

Table 2Incidence of retinal folds in fixed fetal specimens: results from different fixatives

Parameter Bouin’s fluid fixation alone Formalin/Bouin’s fluid fixation Davidson’s/Bouin’s fluid fixation Davidson’s fixation alone

No. (%) of heads examined 54 18a 33 0b

No. (%) of litters examined 13 4a 8 0b

Fold in region of fovea (unilateral/bilateral)No. (%) of heads affected 50 (93) 18 (100) 3 (9) –No. (%) of litters affected 13 (100) 4 (100) 2 (25) –

Retinal fold (non-foveal) (unilateral/bilateral)No. (%) of heads affected 39 (72) 18 (100) 5 (15) –No. (%) of litters affected 13 (100) 4 (100) 2 (25) –

As thfl d.

ossibl

ot

4

dswnTucr

faaasFatet

s(ati

fda

fltpscetp

wbrn

raftti

C

R

[

a Crystalline deposits within the heads examined made sectioning very difficult.uid fixation alone and formalin/Bouin’s fluid fixation no more heads were sectioneb No appreciable decalcification was apparent and, therefore, sectioning was imp

ther areas of the retina was reduced to approximately one-fifth ofhat seen with Bouin’s fluid alone.

. Discussion

In this study, where no test compound was given, a high inci-ence (78%) of slight retinal folds was seen in the Bouin’s fixedamples. The presence of retinal folds in fixed samples, whichere not present in the fresh samples, suggests that “slight reti-al folding” in the near term rabbit fetus is artifactual in origin.his suggestion is further supported by the fact that heads fixedsing a different fixation regime (where Davidson’s fixation pre-eded Bouin’s fluid exposure) showed a much lower incidence ofetinal folding.

Although there was some variability in the number of retinalolds detected by each of the three observers (Table 1), the overallgreement on incidence between the MRI and stereomicroscopynalyses was very good both at the tabulated litter incidence levelnd in observer concordance of the individual fetus (data nothown). Whilst some folds were obvious, such as that shown inig. 3B and C, others were smaller and more subtle. Minor discrep-ncies are probably attributable to the different observer ratherhan the different technique used to reveal the folds, and reflect anlement of the subjectivity involved in observing the finding dueo the subtlety of some folds.

The retinal fold incidence of 78% in the fixed specimens in thistudy is higher than the previously observed range, of up to 45%unpublished background data). Since observers were not blindeds to the purpose of the study it is not possible to say whether thereruly was a higher incidence in this study, or whether the reported

ncidence reflects some observer bias.

The higher than expected incidence of retinal folds in the fixedetuses has allowed this investigation to produce convincing evi-ence, using a relatively small number of animals, that folds inll parts of the retina are most likely a fixation artifact. That the

[

[

ere was no apparent difference in the number of folds observed, between Bouin’s

e.

olds are a Bouin’s fixation artifact could also contribute to thearge variation in the observation across laboratories. Assuminghese fixation artifacts are influenced by the rate of Bouin’s fluidenetration, these fixation artifacts might be influenced by theize of the fetuses, the presence or absence of undisturbed skinovering the head (which may affect the rate of Bouin’s fluid pen-tration) as well as the exact formulation of the Bouin’s fluid andhe temperature at which the fetuses were kept during the fixationeriod.

From the results obtained from the use of different fixatives itould appear that using Davidson’s fixative for 2 weeks, followed

y 1 week in Bouin’s fluid markedly reduces the incidence of “slightetinal folds”. As a result this laboratory has now changed to thisew fixation regime.

In conclusion, there is strong evidence from this study that slightetinal folds in the rabbit fetus are artifactual in origin, and occurs a result of fixation in Bouin’s fluid. The use of Davidson’s fixativeor 2 weeks, followed by 1 week in Bouin’s fluid markedly reducedhe incidence of “slight retinal folds”. Studies of candidate drugshat report folds in fixed retinas from rabbit fetuses should now benterpreted accordingly.

onflict of interest

None.

eferences

1] Michaelides M, Luthert PJ, Cooling R, Firth H, Moore AT. Norrie dis-ease and peripheral venous insufficiency. British Journal of Ophthalmology

2004;88(11):1475.

2] Nishimura M, Yamana T, Sugino M, Yamana Y, Minei M, Sanui H. Falciform retinalfold as sign of familial exudative vitreoretinopathy. Japanese Journal of Ophthal-mology 1983;27:40–53.

3] Niggeschulze A, Weisse I, Notman J. Fixation-induced cyst-like spaces in thebrains of rabbit fetuses. Archives of Toxicology 1977;37:227–32.

2 e Toxi

[

[

66 J. French et al. / Reproductiv

4] Latendresse JR, Warbrittion AR, Creasy DM. Fixation of testes and eyes usinga modified Davidson’s fluid: comparison with Bouin’s fluid and conventionalDavidson’s Fluid. Toxicologic Pathology 2002;30(4):524–33.

5] Addition to: Wise LD, Beck SL, Beltrame D, Beyer BK, Chahoud I, Clark RL, et al.Terminology of developmental abnormalities in common laboratory (version 1).Teratology 1997;55(4):249–92. Addition in preparation.

[

[

cology 26 (2008) 262–266

6] Van Julsingha EB, Bennett CB. A dissecting procedure for the detection of abnor-malities in the rabbit fetal head. Methods in prenatal toxicology. Stuttgart:Neubert D Georg Thieme; 1977.

7] Siegel S, Castellan NJ. Nonparametric statistics for the behavioral sciences. 2nded. London: McGraw-Hill; 1988. p. 75–7.