Postgraduate Medical Journal (October 1975) 51, 716-721.

Small intestinal biopsy in childhood coeliac disease

R. A. RISDON*M.D., M.R.C.Path.

D. G. WADBROOKtB.Sc.

E. A. MEINHARDtM.B., M.R.C.Path.

J. W. KEELING*:M.B., M.R.C.Path.

* The Hospitalfor Sick Children, Great Ormond Street, and t The Royal Free Hospital, London

SummarySmall intestinal biopsies from children with treatedand untreated coeliac disease have been analysedmorphometrically and compared with controls. Twomethods have been used; one, described by Dunnilland Whitehead (1972) provides indices of surface-to-volume ratio and of mucosal volume. The other,described by Meinhard, Wadbrook and Risdon (1965),involves tracing the microscopic image of the biopsyon to computer data cards to produce measured en-coded data for direct analysis with a computer.

Biopsies from untreated coeliacs were clearlydistinguished from control specimens by both tech-niques. Comparison of these two groups by computercard morphometry shows the mucosal lesion incoeliac disease to be associated with a spatial re-distribution of the tissue components rather than achange in their absolute amounts. No difference in thetotal mucosal volume was found so that a true mucosalatrophy does not occur in this condition.

Surface-to-volume (c: lh) ratios were measured intwenty-eight children treated empirically with agluten-free diet for suspected coeliac disease, inbiopsies taken before and after a gluten challenge. Bythis means the diagnosis of coeliac disease was con-firmed in eighteen of these patients.

In morphologically normal biopsies correlation ofc: lh ratios with age showed significantly smallervalues in younger children. Slight changes in biopsiesfrom young children should, therefore, be interpretedcautiously and should not necessarily be regarded aspathological on the evidence of the villous pattern alone.

IntroductionImprovements in biopsy techniques have led to

their increased use both for diagnosis and in follow-ing a disease process or its response to treatment inserial biopsies from the same individual. This isparticularly true of childhood coeliac disease; cur-rent definitions of this condition require the demon-stration of both a mucosal lesion and its response to

t Present address: John Radcliffe Hospital, Oxford.Correspondence: Dr R. A. Risdon, Department of Histo-

pathology, The Hospital for Sick Children, Great OrmondStreet, London WCIN 3JH.

gluten withdrawal or reintroduction in sequentialjejunal biopsies (Townley and Barnes, 1973; Ander-son, Gracey and Burke, 1972). Increasingly, there isa need for quantitative methods for assessing thechanges seen in these specimens objectively.

Previous attempts to measure jejunal biopsychanges have been recently reviewed by Dunnill andWhitehead (1972) and by Chapman et al. (1973). Inthe present study morphometric methods have beenapplied to both single and serial biopsies fromchildren. The technique of Dunnill and Whiteheadcombines point counting and linear interceptprinciples (Chalkley, Cornfield and Park. 1949) tomeasure surface-to-volume ratios and mucosalvolume indices.Computer card morphometry (Meinhard, 1974)

requires the tracing of the microscopic image of thebiopsy on computer data cards. These are printedwith a grid. the cross-points of which form amatrix of 840 points. Quantitative data are recordedas marks where tissue components overlie the cross-points, thus retaining their spatial arrangement with-in the tissue. The data are fed, via an Optical MarkData Card Reader, into a minicomputer for analysis.

Materials and methodsPeroral biopsies of small intestinal mucosa were

obtained under fluoroscopic control using a Crosbycapsule. The specimens were taken immediatelydistal to the ligament of Treitz. The handling andprocessing of the biopsies have been described indetail elsewhere (Risdon and Keeling, 1974).

MorphometryMethod 1 (Dunnill and Whitehead, 1972)

This technique involves superimposing a patternof fifteen lines of equal length (drawn to the speci-fication of Weibel, 1963) on to the image of thebiopsy (Fig. 1), using an eye-piece graticule.Two parameters are counted; 'hits' where theend points of the lines lie over the mucosa, and'cuts' where the lines cross the surface epithelium.The magnification is kept constant, and when countsin each microscopic field are completed the sectionis racked to the next adjoining field. Counting iscontinued until a total of about 200 'hits' have been

by copyright. on O

ctober 6, 2020 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.51.600.716 on 1 O

ctober 1975. Dow

nloaded from

Small intestinal biopsy

ll_ts-~ _.4_'

FIG. 1. Superimposed over a section of a control biopsy is a template of fifteen linesof equal length arranged as specified by Weibel (1963). 'Hits' are counted where theendpoints of the lines lie over the mucosa; 'cuts' are counted where the lines crossthe surface epithelium. At the magnification used, the length (I) of each line was1-4 x 10-2 cm (HE x 17-5).

recorded. The number of microscopic fields used isalso noted. The surface-to-volume ratio index isgiven by c: lh (c is the number of 'cuts', h is thenumber of 'hits' and I the length of the lines). Themucosal volume index is given by h, the averagenumber of 'hits' per microscopic field counted.From ninety children, 113 jejunal biopsies were

studied by this method. Thirty-seven biopsies wereobtained from children investigated for short statureor failure to thrive. Since none had investigationalevidence of malabsorption and all biopsies weresubjectively normal, these specimens were used as'controls'. Thirty biopsies were taken from childrenwith typical coeliac disease before any dietarytreatment. Forty-six were paired biopsies fromtwenty-three children treated empirically for at least2 years with gluten-free diet for suspected coeliacdisease. This has been diagnosed previously on clinicalgrounds alone. One biopsy was taken before, and asecond biopsy was taken some 12 weeks after re-introduction of dietary gluten.

Method 2 (Meinhard et al., 1975)The method requires the tracing of a correctly

orientated microscopic image of the biopsy on tocomputer data cards; the grids printed on the cardsprovide a point counting graticule of 840 samplingpoints.Each drawing is done on two superimposed data

cards; an upper punched master card forms a tem-plate over a lower component card. Each cross-point of the grid printed on the component card isvisible through a punched hole in the master card.After calibration, selected components of the biopsyare traced on the cards using a drawing tube attach-ment to the microscope. This produces a sketch onthe upper master card and a series of marks on thelower component card (Fig. 2). A different com-ponent card is used for each component to bemeasured but the master card is retained. Thus acomposite image of the biopsy is gradually built upon the master card. The component cards are readby an optical mark data card reader. The informationis then fed via a 12-line input into a mini-computerin binary image form (i.e. as an array of l's and 0's).FORTRAN IV programs allow calculation ofsurfacearea and component volumes in absolute (metric)units. Calibration before making the drawings meansthat the distance between lines on the grid is definedand hence an area can be assigned to each samplingpoint, and by summation to each component withinthe matrix. Distances between sampling points,diagonally as well as vertically and horizontally, canbe measured and thus the distance along a contourcan be derived. Similarly, the length of the muscularismucosae can be measured and the point count andupper contour values can be expressed as an area orperimeter measurement per 1 mm of muscularis

717

by copyright. on O

ctober 6, 2020 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.51.600.716 on 1 O

ctober 1975. Dow

nloaded from

R. A. Risdon et al.

(b)

Photomicrograph of the biopsy

The master card

'Not mucoso' component cord

Surface epithelium component card

Crypt epithelium component card

FIG. 2. (a) The progression from a control specimen via the master card to the component cards. (i) Photomicrograph ofthe biopsy; (ii) the master card; (iii) 'not mucosa' component card; (iv) surface epithelium component card; (v) cryptepithelium component. (b) Similar progression for a 'flat' biopsy from a patient with coeliac disease.

TABLE 1. Surface-to-volume ratios and mucosal volume indices in biopsies from controlsand untreated coeliacs

c: lh ratio h

No. Range Mean s.d. Range Mean s.d.

Controls 37 26&3-56 0 43-5 8-1 14 3-23 7 18'7 2-2Untreated

coeliacs 30 4 7-14 3 9-7 2-4 14 8-23 3 19 6 2-3

mucosae. Volume proportions and surface-to-volume ratios can also be derived relative to thetotal mucosal volume.

In this paper some of the results obtained withthis method in biopsies from the controls and intwenty-six of the specimens from untreated coeliacsare considered. A fuller and more extended study isreported elsewhere (Meinhard et al., 1975; Wad-brook and Meinhard, 1975).

ResultsMethod 1

Results for the biopsies from controls and un-treated coeliacs are set out in Table 1 and Fig. 3.

The numerical values for c: Ih ratios and hindices are different from those previously reported(Risdon and Keeling, 1974), because of changes inmethodology. Previously, 'hits' were scored onlywhen the ends of lines fell on the lamina propria ormucosal glands, thus excluding surface epitheliumfrom volume determinations. Experience with com-puter card morphometry has shown that in normalspecimens surface epithelium comprises some 30%.of the total mucosal volume. Differences we pre-viously reported in mucosal volume between con-trols and 'flat' biopsies have been shown to bespurious, and are due to the exclusion of surfaceepithelium when counting 'hits'.

718

by copyright. on O

ctober 6, 2020 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.51.600.716 on 1 O

ctober 1975. Dow

nloaded from

Small intestinal biopsy 719

12

0

06

4

2

25 30 35 40 45 50 55 60c: Ih rotio

FIG. 3. A histogram showing the normal distribution ofc: lh ratios in the thrity-seven control biopsies. Themean age of the 11 cases with c : h ratios below 40(corresponding to the shaded area) was 29-2 months,whilst that of the remaining twenty-six cases was69-3 months. The difference is significant (t = 2-78,P < 0-01).

60 -

40 0

40 --

20

Pregluten PostglutenFIG. 4. The c : h ratios in paired biopsies from patientsbefore and after the reintroduction of dietary gluten.The closed circles represent morphologically normalbiopsies and the open circles abnormal biopsies. Biopsiesfrom the same patients are joined by lines. Means andtwice the standard deviations for controls and biopsiesfrom untreated coeliacs are indicated.

5

I

0o

E

E 2 io

a2

0

Controls Untreatedcoeliacs

FIG. 5. The surface areas of controls and biopsies fromuntreated coeliacs. Means and standard deviations areindicated.

*

a, 026

o

E

") w

.' 010- T

Eo

E 0.14-

. -

-50,

'S 0-06

E

0-02-

Controls Untreatedcoeliacs

FIG. 6. Volumes of surface epithelial cells in controlsand biopsies from untreated coeliacs.

by copyright. on O

ctober 6, 2020 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.51.600.716 on 1 O

ctober 1975. Dow

nloaded from

720 R. A. Risdon et al.

0.70-

o0

°, 0.60- ·o 0E

E

E

E·

o

i0'30-E

| 00-"I

o

E

2 0-20o

E 0-10-

O

00

Controls Untreatedcoel iacs

FIG. 7. The total mucosal volumes in controls and un-treated coeliacs.

Results from the paired biopsies are shown inFig. 4. A careful dietary history showed that allbut three of the twenty-three children consideredhad kept rigidly to a gluten-free diet; first biopsiesin these twenty patients had c : lh ratios well withinthe control range. Of the first biopsies obtained fromthe three children in whom dietary lapses occurred,c: lh ratios were well below the control range intwo, and just at its lower end in the third. Thereintroduction of dietary gluten produced a markedfall in c : lh ratios in thirteen of the twenty childrenwith normal values on a gluten free diet; in eight ofthem c : Ih ratios were similar to those in untreatedcoeliacs and in the other five values fell in an inter-mediate range corresponding to so-called 'partialvillous atrophy'. In seven patients no fall in c: lhratio occurred when gluten was reintroduced.

Method 2Numerous differences were recorded between

controls and untreated coeliacs. Figures 5 and 6show the results for both groups for surface areaand surface epithelial cell volume. Total mucosalvolumes in the two groups (Fig. 7) did not differ(t = 1.19; P > 02).DiscussionMorphometric methods used in histopathology

(Dunnill, 1968; Weibel, 1963) involve the prediction

of three dimensional volume and surface areameasurements of structures randomly distributedwithin an organ or tissue, from area or perimetermeasurements of those structures seen in two dimen-sional sections. However, the components of tissuessuch as jejunal mucosa are not randomly distributed,so that meaningful comparisons between specimensrequire that they be prepared in a similar fashion.In particular, a standard orientation of the tissuesections is chosen to best reveal their structure,which in jejunal mucosa is at right angles to itssurface. The level in the small intestine from whicha biopsy is taken is also important since variations inthe mucosal appearances may occur at different sites(Rubin and Dobbins, 1965).Of the two methods of quantitative analysis

employed in the present study, that of Dunnill andWhitehead (1972) provides a simple direct techniqueneeding a minimum of special equipment (an eye-piece graticule specially made by Graticules Ltd);its advantages over other methods employing directmeasurements of mucosal components is discussedby Dunnill and Whitehead (1972).Computer card morphometry (Meinhard, 1974) is

a modification of point counting (Glasgolev, 1933)but has a number ofadvantages over more traditionaltechniques. The data are already coded for computeranalysis which forms an integral part of the technique.Furthermore, since the spatial arrangement of thepoint counts are retained, the measured image can,in effect, be reconstructed within the computer. It isin fact possible to produce pictorial as well as numeri-cal printouts from the data suitable for inclusion inthe patients' notes (Meinhard et al., 1975). Sincethe matrix is calibrated, volume and surface areadeterminations can be derived in absolute terms.The practical value of quantitative analysis of

jejunal biopsy specimens, which compared withpurely subjective interpretation is both tedious andtime consuming, lies only partly in distinguishingand defining minor degrees of villous abnormality.It also allows objective comparisons between speci-mens from different individuals or between serialbiopsies from the same subject.By determining surface-to-volume (c: lh) ratios

in biopsies taken both before and after a glutenchallenge in twenty-three children previously treatedempirically with a gluten free diet, the diagnosis ofcoeliac disease was confirmed in sixteen. In thirteen asignificant fall in c: lh ratios occurred as a resultof gluten ingestion confirming its deleterious effecton mucosal morphology in these patients; pre-challenge values were well within the control rangeshowing that in children with coeliac disease themucosal lesion can heal completely with glutenwithdrawal. In seven patients both pre- and post-challenge biopsies were normal. These children

by copyright. on O

ctober 6, 2020 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.51.600.716 on 1 O

ctober 1975. Dow

nloaded from

Small intestinal biopsy 721

almost certainly do not have coeliac disease, althougha further biopsy after two years of gluten ingestionmay be necessary to exclude a 'delayed' mucosalresponse to gluten (Walker-Smith, 1970, 1972).A point of some interest regarding the morpho-

genesis of the mucosal lesion in coeliac disease isapparent from the computer card results. Althoughthe surface epithelial volume is lower and the cryptepithelial volume greater in the biopsies from un-treated coeliacs compared to the controls, the totalmucosal volumes are the same (Fig. 8). Further, thelamina propria volume is only slightly greater in thecoeliac biopsies despite an obvious increase ininflammatory cell infiltration. These results implythat the mucosal lesion involves a redistribution oftissue components, particularly surface and cryptepithelium, and not a change in the absolute amounts:that is, there is no true mucosal atrophy.

Surface ::: ..::::epithelium :...: ..:.:....... :: .. : :..

:::::::::::::::::::::::::::::::::::-:: .................. ....... ..I.......

...'''''':''' '::':''' ,. ':':.'' ~''' `':' ''' ' ''' ''...p

.. .: .. ':'

Contro ls U ntreatedcoel iacs

..

....................................'.. ..'

'....

FIG. 8. Comparison of the mean volumes of totalmucosa, epithelium and lamina propria in controls anduntreated coeliacs. In the coeliac biopsies the fall insurface epithelial volume is associated with a rise in cryptepithelial volume.

Analysis of surface-to-volume (c : lh) ratios in thecontrol biopsies relating them to the patients' agesshowed that the lower c : lh ratios tended to occurin younger children (Fig. 3). This suggests that thecomplexity of the villous pattern increases at avarying rate during early childhood so that someyoung children have a relatively 'immature' mucosalmorphology. This is supported by consideration ofthe dissecting microscope appearances of jejunalbiopsies from young children which often have a'ridged' pattern which might be considered patho-logical in an adult.

ReferencesANDERSON, C.M., GRACEY, M. & BURKE, V. (1972) Coeliac

disease. Some still controversial aspects. Archives ofDisease in Childhood, 47, 292.

CHALKLEY, H.W., CORNFIELD, J. & PARK, H. (1949) A methodfor estimating volume-surface ratios. Science, 110, 295.

CHAPMAN, B.L., HENRY, K., PAICE, F., STEWART, J.S. &COGHILL, N.F. (1973) A new technique for examiningintestinal biopsies. Gut, 14, 903.

DUNNILL, M.S. & WHITEHEAD, R. (1972) A method for thequantitation of small intestinal biopsy specimens. Journalof Clinical Pathology, 25, 243.

DUNNILL, M.S. (1968) Quantitative methods in histology.In: Recent Advances in Clinical Pathology; Series V(Ed. by S. C. Dykes) p. 40. Churchill, London.

GLASGOLEV, A.A. (1934) Quantitative analysis with themicroscope by the 'point' method. Engineering andMineralogy Journal, 135, 399.

MEINHARD, E.A. (1974) Histoquantitation using computerdata cards. Journal of Microscopy, 101, 95.

MEINHARD, E.A., WADBROOK, D.G. & RISDON, R.A. (1975)Computer card morphometry of jejunal biopsies in child-hood coeliac disease. Journal of Clinical Pathology (inpress).

RISDON, R.A. & KEELING, J.W. (1974) Quantitation of thehistological changes found in small intestinal biopsy speci-mens from children with suspected coeliac disease. Gut,15, 9.

RUBIN, C.E. & DOBBINS, W.O. (1965) Peroral biopsy of thesmall intestine: a review of its diagnostic usefulness.Gastroenterology, 49, 676.

TOWNLEY, R.R.W. & BARNES, G.L. (1973) Intestinal biopsyin childhood. Archives of Disease in Childhood, 48, 480.

WADBROOK, D.G. & MEINHARD, E.A. (1975) (in prepara-tion).

WALKER-SMITH, J.A. (1970) Transient gluten intolerance.Archives of Disease in Childhood, 45, 523.

WALKER-SMITH, J.A. (1972) Transient gluten intolerance(correspondence). Archives of Disease in Childhood, 47,155.

WEIBEL, E.R. (1963) Principles and methods for the morpho-metric study of the lung and other organs. LaboratoryInvestigation, 12, 131.

by copyright. on O

ctober 6, 2020 by guest. Protected

http://pmj.bm

j.com/

Postgrad M

ed J: first published as 10.1136/pgmj.51.600.716 on 1 O

ctober 1975. Dow

nloaded from