-
Gut 1993; 34: 68-74
Quantification of disease activity in Crohn's diseaseby computer
analysis of Tc-99m.hexamethylpropylene amine oxime (HMPAO) labelled
leucocyteimages
M H Giaffer, W B Tindale, S Senior, D C Barber, C D
Holdsworth
AbstractThe reliability and clinical applications ofcomputerised
image analysis measurement ofbowel uptake of Tc-99m HMPAO
labelledleucocytes has been examined as a measure ofdisease
activity in Crohn's disease. In 54studies carried out on 33
patients with estab-lished Crohn's disease, the mean 'scan score',a
quantitative assessment of image intensity,was 82-1 SEM (13.6), in
patients with clinicailyactive disease compared to 24-7 (7.0) in
thosewith quiescent disease, p.O-0005. A signifi-cant correlation
was found between the scanscore and Crohn's Disease Activity Index
(rs=0-52, p
-
Quantification ofdisease activity in Crohn's disease by computer
analysis oflabelled leucocyte images
DISEASE ACTIVITY
Disease activity was assessed using: (A) Crohn'sDisease Activity
Index - CDAI,' the diseasebeing considered active if the CDAI value
wasgreater than 150. (B) Simple Index of Harveyand Bradshaw - SI.2.
A value of -4 denotesactive disease. (C) Laboratory tests
reflectingactive gut inflammation,9 these included:haemoglobin:
normal: male :-12-5 g/dl, female-11 5 g/dl; platelet count: normal:
150-400x109/1; erythrocyte sedimentation rate: normal:-20 mm in
first hour; serum albumin: normal:-35 g/l; C-reactive protein:
normal:
-
Giaffer, Tindale, Senior, Barber, Holdsworth
marrow; in some cases, urinary bladder activityalso
appeared.Bowel uptake was evident as early as 40
minutes. Some hepatobiliary excretion wasevident on some of the
four hour images, asnoted by previous workers.2' Compared
withIndium-ill, the image quality of Tc-99mHMPAO scans was
superior.
Reproducibility studies on the data used tocalculate the scan
score yielded a coefficient ofvariation of 5*5% .
DISEASE ACTIVITYFigure 2 shows that patients with active
Crohn'sdisease have a significantly higher scan score(mean 82-1
(13-6), median 51) than those withquiescent disease (mean 24-7
(7T0), median 15)(p.O-005). Optimum separation between thegroups
was attained with a threshold scan scoreof 20. On this basis, the
technique had a sensi-tivity of 82%, a specificity of 77%, a
predictivevalue of 91%, and a false positive rate of
8-8%.Discordance between the CDAI and the scanscore was observed on
10 occasions. In seven ofthese, raised CDAI >150 was associated
withrelatively low scan score (s20). In each of thesepatients it
was clear that the raised CDAI wascaused either by the presence of
symptomswhich although contributing substantially to the
400 -
300 -
Figure 1: (A) Computergenerated 'normal' imagecontaining
backgroundactivity only - that is, bonemarrow, liver, and
spleen.(B) Patient's image, scaledto fit the normal
image,containing, in addition tobackground activity, someabnormal
bowel uptake inthe terminal ileum. (C)Abnormal bowel
uptakeisolatedfrom backgroundactivity by subtracting thenormalfrom
the patient'simage. The scan score is thecount in abnormal
bowelarea adjusted to the injected99MTc dose.
man's coefficientability.
200 -
.
0
S
S
0
0
0
*0
00
> 150 S 150(n = 38) (n = 13)
Crohn's disease activity index
a)
oQcJcoC)
U)
100 -
and p, the associated prob-
0-
ResultsThe viability of the injected leucocytes wasestimated at
95% using the standard trypan blueexclusion test. After injection
of the labelledleucocytes, radioactivity on the abdominal scanwas
distributed so that spleen>liver>bone
Figure 2: Scan score in patients with active and
quiescentCrohn's disease. A threshold of20 (shaded area) separates
thetwo groups with the highest predictive value and
overallaccuracy.
A
B
I
C
ZI Medianl
70
on June 19, 2021 by guest. Protected by copyright.
http://gut.bmj.com
/G
ut: first published as 10.1136/gut.34.1.68 on 1 January 1993.
Dow
nloaded from
http://gut.bmj.com/
-
Quantification ofdisease activity in Crohn's disease by computer
analysis ofoflabelled leucocyte images
400 -
350 -
300 -
250-aD0
X 200-c
150-
100-
50 - * .* .// .
0 ./ .0^ SS...
100 200 300 400 500 600Crohn's disease activity index
Figure 3: Correlation between scan score and Crohn's
DiseaseActivity Index.
CDAI were not caused by active inflammation -for example,
superimposed symptoms of irrit-able bowel syndrome in three
patients andsubacute intestinal obstruction from fibrousstricture
in two patients, or to symptomaticanaemia (one case). One patient
with Crohn'scolitis had CDAI of 376 (main symptoms werediarrhoea
and feeling unwell) and scan score of13 1 but only 'mild'
inflammatory changes weredemonstrated endoscopically and
histologically.On the other hand, a normal CDAI ofless than
150 but a high scan score indicating activeCrohn's disease was
recorded in three scans.Two of these were from one patient with
newlydiagnosed and radiologically active terminal ilealCrohn's
disease who was treated with elementaldiet for four weeks. An
initial scan score of 77 9decreased to 29-9 after dietary
treatment. Thethird scan came from a patient with Crohn'scolitis
who despite the absence of symptomscontinued to have persistently
abnormal labora-tory tests and a scan score of 80-6. Clearly
inthese three patients the abnormal scan score was
a more accurate indicator of the degree of activitythan the
CDAI.When disease activity was assessed using the
visual grading method by two independentobservers, there was
complete agreement in only26% of scans. A difference of at least
two gradeswas recorded in 30% of cases. The visual gradingscore was
high indicating active disease in nine(45%) of the 20 scans taken
from patients withclinically quiescent disease and who had a
low(-
-
Giaffer, Tindale, Senior, Barber, Holdsworth
I rs=0-57 p 0-001
*0
0*0
0* 0:0
0-5 1 0 1 5 20 25 30Acid alpha glycoprotein (g/l)
20 40 60 80 100 120 140 160
C-reactive protein (mg/I)
Figure 5: Correlation between scan score and C-reactive
protein.
of radiolabelled leucocytes to localise andestimate disease
activity in inflammatory boweldisease provides an objective
technique for theassessment of disease activity. Of the
radio-isotopes which have been used for labellingleucocytes,
Indium-ill is the most widelystudied. 1012 17 Its long half life
(67 hours) permitsfaecal excretion studies, which are considered
bysome as the 'gold standard' parameter of diseaseactivity in
Crohn's disease. 12 The need for collec-tion of stools over four
days, and a radiation doselimiting repeated studies are the main
dis-advantages of faecal indium excretion studies.Tc-99m HMPAO,
unlike other Tc-99m
agents, is relatively stable in granulocytes com-pared with
other blood cells and can therefore beused as a selective
granulocyte label without theneed for the tedious technique
required for theseparation of these cells from mixed
leucocytesuspensions. From the point of view of radiationdosimetry,
image quality, availability andrelative expense Tc-99m is
preferable to Indium-111. The presence of non-specific biliary
excre-tion and short half life, however, render faecalcollection
studies after Tc-HMPAO labelledleucocytes of little clinical use.
Disease activitymust be estimated from the bowel
scintigramsinstead. The only method hitherto available forthe
quantification ofbowel activity on abdominalscans has been the
grading of the bowel uptakeby visually comparing it with that of
the spleen,liver, and bone marrow." 1617 This method is
assubjective as the clinical indices and dependsupon the presence
of normal spleen, liver, and
Figure 6: Correlation between scan score and alpha
acidglycoprotein.
bone marrow. Variations in the uptake oflabelled leucocytes in
these organs have beenreported2' and patients with inflammatory
boweldisease may have associated hyposplenism23which may affect the
splenic uptake. In calculat-ing the scan score, it is essential to
separate theabnormal bowel uptake from the backgroundactivity.
Without such separation disease activitywill frequently be
overestimated particularlywhen the disease is quiescent (data not
shown).Although not the objective of this study, we havecompared
our method with that in which diseaseactivity was estimated by
comparing the boweluptake with that ofthe bone marrow, spleen,
andliver.We found poor correlation between the two
methods with the visual grading method having asignificant
interobserver variation. A significantproportion of our patients
who had clinicallyquiescent disease and who had a low scan
scorewere classed as having active disease by the visualgrading
method. In patients with active disease,the visual score tended to
under estimate diseaseactivity.The computer based analysis of bowel
images
described in this paper meets most of the criteriafor an ideal
activity marker activity in Crohn'sdisease. The scan score is
objective, and reliablydifferentiates between active and
inactiveCrohn's disease. The seven patients withspuriously raised
CDAI as a result of non-inflammatory conditions or complications
ofCrohn's disease, such as fibrous strictures, werecorrectly
identified by the scan score. We foundthe scan score particularly
helpful in the accurateestimation of disease activity in these
circum-
TABLE Correlation matrix betzveen scan score and laboratory
tests in patients with Crohn's disease
Small bowel disease Large bowel disease All sites
rs p rs p rs p
Haemoglobin -0-58 0 01 -0-879 0 001 -0-66 0 0001Erythrocyte
sedimentation rate 0-2 NS 0 4 NS 0-22 NSPlatelet count 0 53 0-05
0-92 0-0006 0-47 0-006Serum albumin -0-8 0 0057 -0-81 0-008 -0-61 0
0001C-reactiveprotein 0 46 NS 0-711 0-03 0-72 0 0001Alpha-acid
glycoprotein 0-8 0 0059 0-56 0 09 0-57 0 001
rs = Spearman's correlation coefficientp = Associated
probabilityNS = Not significant
300
250 -
200 -
o0)
uc 150-0
uz
100 -
50 -
300
250
0
0 0
. 00
200a)0oC' 150 -
cn
0
0
0
0
S
0
0
50
0 I
72
.
.
.
on June 19, 2021 by guest. Protected by copyright.
http://gut.bmj.com
/G
ut: first published as 10.1136/gut.34.1.68 on 1 January 1993.
Dow
nloaded from
http://gut.bmj.com/
-
Quantification ofdisease activity in Crohn's disease by computer
analysis oflabelled leucocyte images 73
stances, often enabling unnecessary therapeuticmanipulations to
be avoided. Conversely ourthree patients with a normal CDAI value
butraised scan score all had some other confirmatoryevidence of
active gut inflammation such aspersistently abnormal laboratory
tests or palp-able abdominal mass. This shows that theclinical
indices can underestimate diseaseactivity, this being particularly
likely when thereare minimal abdominal symptoms or when
pre-sentation is with features such as weight loss.The scan score
correlated with all laboratory
tests generally accepted to reflect active gutinflammation22
except for the erythrocyte sedi-mentation rate. Although the
erythrocyte sedi-mentation rate has been used to measure
diseaseactivity in inflammatory bowel disease34 and topredict
relapse after treatment,24 it is frequentlynormal in some patients
with unequivocallyactive disease.24 25 Our data support the results
ofother studies which have found the erythrocytesedimentation rate
to be of very limited value.Powell-Tuck26 found no significant
correlationbetween the erythrocyte sedimentation rate
andsigmoidoscopic appearance in patients withulcerative colitis and
in Crohn's disease. Cookeand Prior25 showed that serum C-reactive
proteinand albumin but not the erythrocyte sedimenta-tion rate
correlated with disease activity. Thescan score correlated well
with C-reactive proteinand alpha-I acid glycoprotein. Both are
con-sidered by some as the most reliable markers ofdisease activity
in Crohn's disease.4 272 Althoughhaemoglobin concentration and
serum albuminmay be affected by factors other than active
gutinflammation such as malnutrition and bleeding,hypoalbuminaemia,
and anaemia are frequentfeatures of active Crohn's disease.2829 The
scanscore correlated with both albumin and haemo-globin. It is of
interest that the correlationbetween the scan score and albumin is
highest inpatients with small bowel involvement.One of the main
applications of the scan score
is in difficult cases when accurate estimation ofdisease
activity will have important therapeuticimplications. Laboratory
tests are frequentlynormal in patients with unequivocally
activedisease and vice versa.22 Symptoms may thereforebe
erroneously attributed to other non-inflammatory conditions. The
scan score isparticularly useful in these circumstances forits
sensitivity approaches 100%. Conversely,patients with quiescent
disease who continue tohave persistently abnormal blood tests have
ahigh recurrence rate.332 The scan score showedthat this is
probably caused by the presence ofsubclinical active gut
inflammation for all ofthese patients had raised scan score. A
furtherimportant value of the scan score is in serialstudies in
individuals'particularly in the contextof therapeutic trials. We
have recently shownthat in patients with Crohn's disease treated
withelemental diet remission can be defined moreobjectively by the
use of the scan.3
In summary, the computer based analysis ofbowel images obtained
from patients withCrohn's disease meets most of the criteria foran
ideal marker of disease activity. Tc-99mHMPAO is widely available,
delivers a relativelysmall radiation dose, has a short half life
and
because of its emission characteristics producespictures of
superior quality to those obtainedwith Indium-Ill. The calculation
of the scanscore from these images provides a reliable andobjective
indicator of disease activity. Thisindicator can reliably
differentiate betweenactive and quiescent disease and is
particularlyuseful in difficult cases when symptoms may notreflect
active disease or when severe symptomsare not backed by abnormal
laboratory tests. Thescan score correlates well with widely
usedclinical and laboratory markers of diseaseactivity. Of these,
correlations with the serumalbumin, and alpha-I acid glycoprotein
particu-larly in small bowel disease, and C-reactiveprotein,
particularly in large bowel disease, werethe most significant. The
complete lack of cor-relation with the erythrocyte sedimentation
ratereinforces the finding of other studies showingthat this
laboratory test may not be as usefulas once thought in the
assessment of Crohn'sdisease activity, whereas the value of
raisedplatelet count was confirmed, particularly incolonic disease.
The method is particularly valu-able as an objective measurement to
assess theresponse to therapy.
This work was supported by a grant from Cow &
Gate/Nutriciaand the Trustees of the Former United Sheffield
Hospitals.
1 Best WR, Becktel JM, Singleton JW, Kern F. Development ofa
Crohn's disease activity index. National Co-operativeCrohn's
Disease Study. Gastroenterology 1976; 70: 439-44.
2 Harvey RF, Bradshaw JM. A simple index of Crohn's
diseaseactivity. Lancet 1980; i: 514.
3 Van Hees PAM, Van Elteren PM, Van Lier HJ, Van TongerenJHM. An
index of inflammatory activity in patients withCrohn's disease. Gut
1980; 21: 279-86.
4 Andre C, Descos L, Landias P, Fermanian J. Assessment
ofappropriate laboratory measurements to supplement theCrohn's
disease activity index. Gut 1981; 22: 571-4.
5 Fagan EA, Dyck RF, Maton PN, Hodgson HJ, Chadwick VS,Petrie A,
et al. Serum levels of C-reactive protein in Crohn'sdisease and
ulcerative colitis. Eur f Clin Invest 1982; 12:351-9.
6 Harries AD, Fitzsimons E, Fifield R, Dew MJ, Rhodes J.Platelet
count: a simple measure of activity in Crohn'sdisease. BMJ 1983;
286: 1476.
7 Talstad I, Gjone E. The disease activity of ulcerative colitis
andCrohn's disease. Scandj7 Gastroenterol 1976; 11: 403-8.
8 Dearing WH, McGuckin WF, Elueback LR. Serum Alpha-I-acid
glycoprotein in chronic ulcerative colitis. Gastro-enterology 1969;
56: 295-303.
9 Cooke WT, Fowler DI, Cox EV, Gaddie R, Meynell MJ. Theclinical
significance of seromucoids in regional ileitis andulcerative
colitis. Gastroenterology 1958; 34: 910-9.
10 Saverymuttu SH, Peters AM, Hodgson HJ, Chadwick VS,Lavender
JP. Indium-111 autologus leucocyte scanning:comparison with
radiology for imaging the colon in inflam-matory bowel disease. BMJ
1982; 285: 255-7.
11 Saverymuttu SH, Camilleri M, Rees H, Lavender JP,Hodgson HJF,
Chadwick VS. Indium-111 granulocytescanning in the assessment of
disease extent and diseaseactivity in inflammatory bowel disease.
Gastroenterologv1986; 90: 1121-8.
12 Saverymuttu SH, Peters AM, Lavender JP, Pepys MB,Hodgson HJF,
Chadwick VS. Quantitative faecal Indium-111 labelled leukocyte
excretion in the assessment of diseasein Crohn's disease.
Gastroenterology 1983; 85: 1333-9.
13 Ell PJ, Jarritt PH, Cullum I, Hocknell JML, Costa DC, LuiD,
et al. Regional cerebral blood flow mapping with a
new99Tcm-labelled compound. Lancet 1985; ii: 50-1.
14 Peters AM, Danpure HJ, Osman S, Hawker RJ, HendersonBL,
Hodgson HJ, et al. Clinical experience with 99Tc"'hexamethyl
propylene amine oxime for labelling leucocytesand imaging
inflammation. Lancet 1986; ii: 946-9.
15 Schumichen C, Scholmerich EJ, Freiburg B. Tc-99mHMPAO
labelling of leucocvtes for detection of inflamma-tory bowel
disease. Nunci Comnp 1986; 17: 27-16.
16 Scholmerich J, Schmidt E, Schumichen C, Billmann P,Schmidt H,
Gerok W. Scintigraphic assessment of bowelinvolvement and disease
activity in Crohn's disease usingTechnetium 99m-Hexamethyl
prtpylene amine oxime asleukocyte label. Gastroenterologv 1988; 95:
1287-93.
17 Stein DT, Gray GM, Gregorv PB, Andetrson M, Goodvwin
DA,McDougall IR. Location and activity of ulcerative andCrohn's
colitis by IndSium-l 1 1 leukocvte scan. Gwassrfo-enterologv 1983;
84: 388-93.
18 Danpurc HJ, Osman S, Carroll MJ. The development of a
on June 19, 2021 by guest. Protected by copyright.
http://gut.bmj.com
/G
ut: first published as 10.1136/gut.34.1.68 on 1 January 1993.
Dow
nloaded from
http://gut.bmj.com/
-
74 Giaffer, Tindale, Senior, Barber, Holdsworth
clinical protocol for the radiolabelling of mixed leucocyteswith
99mTc Hexamethyl propylene amine oxime. Nucl MedCommun 1988; 9:
465-75.
19 Saverymuttu SH, Peters AM, Danpure HJ? Reavy HJ, OsmanS,
Lavender JP. Lung transit of "Indium labelledgranulocytes.
Relationship to labelling techniques. ScandjHaematol 1983; 80:
151-60.
20 Tindale WB, Barber DC, Giaffer MH, Senior S, HoldsworthCD.
99m-Tc HMPAO labelled leucocyte imaging in Crohn'sdisease: a
subtraction technique for the quantification ofdisease activity.
Clin Phys Physiol Meas 1992; 13: 37-50.
21 Peters AM, Roddie ME, Danpure HJ, Osman S,Zacharopoulos GP,
George P, et al. 99Tcm-HMPAO labelledleucocytes: comparison with 1
'1In-tropolonate labelledgranulocytes. Nucl Med Commun 1988; 9:
449-63.
22 Bartholomeusz FDL, Shearman DJC. Measurement ofactivity in
Crohn's disease. .7 Gastroenterol Hepatol 1989; 4:81-94.
23 Palmer KR, Sherriff SB, Holdsworth CD, Ryan FP.
Furtherexperience of hyposplenism in inflammatory bowel
disease.QJMed 1981; 200: 463-71.
24 Brignola C, Campieri M, Bazzocchi C, Farruggia P, TragnoneA,
Lanfranchi G. A laboratory index for predicting relapsein
asymptomatic patients with Crohn's disease. Gastro-enterology 1986;
91: 1490-4.
25 Whittington PF, Verdain Barnes H, Bayless TM.
Medicalnmanagement of Crohn's disease in adolescence.
Gastro-enterology 1977; 72: 1338-44.
26 Cooke WT, Prior P. Determining disease activity in
inflamma-tory bowel disease. 7 Clin Gastroenterol 1984; 6:
17-25.
27 Powell-Tuck J, Day DW, Buckell NA, Wadesworth J,Lennard-Jones
JE. Correlations between defined sigmoido-scopic appearances and
other measures of disease activity inulcerative colitis. Dig Dis
Sci 1982; 27: 533-7.
28 Fagan EA, Dyck RF, Maton PN, Hodgson HJ, Chadwick VS,Petrie
A, et al. Serum levels of C-reactive protein in Crohn'sdisease and
ulcerative colitis. Eur J7 Clin Invest 1982; 12:351-9.
29 Cooke WT, Fowler DI, Cox EV, Gaddie R, Meynell MJ.
Theclinical significance of seromucoids in regional ileitis
andulcerative colitis. Gastroenterology 1958; 34: 910-9.
30 Kaufman S, Chalmer B, Heilman R, Beeken WA. A prospec-tive
study of the course of Crohn's disease. Dig Dis Sci 1979;24:
269-76.
31 Andre C, Descos L, Vignal J. C-Reactive protein level
predictsCrohn's disease relapse in quiescent patients. Gut 1979;
20:A940.
32 Boirivant M, Leoni M, Tariciotti D, Fais S, Squarcia
0,Pallone F. The clinical significance of serum C-reactiveprotein
levels in Crohn's disease. Results of a prospectivelongitudinal
study. J Clin Gastroenterol 1988; 10: 401-5.
33 Giaffer MH, Tindale WB, Barber D, Holdsworth CD.Definition of
remission in Crohn's disease using a computercontrolled estimation
of bowel uptake of Tc-99mHexamethyl Propylene Amine Oxime (HMPAO)
leucocytebowel scanning. Gut 1990; 31: A1187.
on June 19, 2021 by guest. Protected by copyright.
http://gut.bmj.com
/G
ut: first published as 10.1136/gut.34.1.68 on 1 January 1993.
Dow
nloaded from
http://gut.bmj.com/
