haema ologica | 2018; 103(11) 1873
Recei ed: Janua y 6 2018.
Accep ed: June 26, 2018.
P e-published: July 5, 2018.
©2018 Fe a a S o i Founda ion
Ma e ial published in Haema ologica is co e ed by copy igh .
All igh s a e ese ed o he Fe a a S o i Founda ion. Use o
published ma e ial is allowed unde he ollowing e ms and
condi ions:
h ps://c ea i ecommons.o g/licenses/by-nc/4.0/legalcode.
Copies o published ma e ial a e allowed o pe sonal o in e -
nal use. Sha ing published ma e ial o non-comme cial pu -
poses is subjec o he ollowing condi ions:
h ps://c ea i ecommons.o g/licenses/by-nc/4.0/legalcode,
sec . 3. Rep oducing and sha ing published ma e ial o com-
me cial pu poses is no allowed wi hou pe mission in w i ing
om he publishe .
Co espondence:
a e.nikkila@u a. i
Fe a a S o i
Founda ion
Haema ologica 2018
Volume 103(11):1873-1880
ARTICLE
Acu e Lymphoblas ic Leukemia
doi:10.3324/haema ol.2018.187716
Check he online e sion o he mos upda ed
in o ma ion on his a icle, online supplemen s,
and in o ma ion on au ho ship & disclosu es:
www.haema ologica.o g/con en /103/11/1873
The only well-es ablished isk ac o s o childhood leukemia a e
high-dose ionizing adia ion and Down synd ome.
Compu e ized omog aphy is a common sou ce o low-dose
adia ion. In his s udy, we examined he magni ude o he isk o
childhood leukemia a e pedia ic compu ed omog aphy examina-
ions. We e alua ed he associa ion o compu ed omog aphy scans
wi h isk o childhood leukemia in a na ionwide egis e -based case-
con ol s udy. Cases (n=1,093) we e iden i ied om he popula ion-
based Finnish Cance Regis y and h ee con ols, ma ched by gende
and age, we e andomly selec ed o each case om he Popula ion
Regis y. In o ma ion was also ob ained on bi h weigh , ma e nal
smoking, pa en al socioeconomic s a us and backg ound gamma adi-
a ion. Da a on compu ed omog aphy scans we e collec ed om he
en la ges hospi als in Finland, co e ing app oxima ely 87% o all
pedia ic compu ed omog aphy scans. Red bone ma ow doses we e
es ima ed wi h NCICT dose calcula ion so wa e. The da a we e ana-
lyzed using exac condi ional logis ic eg ession analysis. A o al o 15
cases (1.4%) and en con ols (0.3%) had unde gone one o mo e com-
pu ed omog aphy scans, excluding a 2-yea la ency pe iod. Fo one o
mo e compu ed omog aphy scans, we obse ed an odds a io o 2.82
(95% con idence in e al: 1.05 – 7.56). Cumula i e ed bone ma ow
dose om compu ed omog aphy scans showed an excess odds a io
o 0.13 (95% con idence in e al: 0.02 – 0.26) pe mGy. Ou esul s a e
consis en wi h he no ion ha e en low doses o ionizing adia ion
obse ably inc ease he isk o childhood leukemia. Howe e , he
obse ed isk es ima es a e somewha highe han hose in ea lie
s udies, p obably due o andom e o , al hough unknown p edispos-
ing ac o s canno be uled ou .
In oduc ion
Leukemia is he mos common childhood malignancy.1The incidence a es o
childhood leukemia in Finland a e compa able o hose in o he Eu opean coun-
ies and show a sligh inc easing end up o he 1990s.2Acu e lymphoblas ic
leukemia accoun s o app oxima ely 85% o all childhood leukemias. The majo
his ological sub ype o acu e lymphoblas ic leukemia is p ecu so B-cell acu e lym-
phoblas ic leukemia (~85%).1
Well-es ablished isk ac o s o childhood leukemia include high doses o ion-
izing adia ion, alkyla ing chemo he apy agen s, as well as Down synd ome and
some a e congeni al synd omes such as Fanconi anemia, Bloom synd ome and
a axia elangiec asia.1,3-6 A numbe o gene ic a ian s ha e also been associa ed
Radia ion exposu e om compu e ized
omog aphy and isk o childhood leukemia:
Finnish egis e -based case-con ol s udy
o childhood leukemia (FRECCLE)
A e Nikkilä,1Jani Rai anen,2,3 Olli Lohi1,4 and Anssi Au inen2,3,5
1Facul y o Medicine and Biosciences, Uni e si y o Tampe e; 2Facul y o Social
Sciences, Uni e si y o Tampe e; 3UKK Ins i u e o Heal h P omo ion Resea ch,
Tampe e; 4Tampe e Cen e o Child Heal h Resea ch, Uni e si y o Tampe e and
Tampe e Uni e si y Hospi al and 5STUK – Radia ion and Nuclea Sa e y Au ho i y,
Helsinki, Finland
ABSTRACT
wi h inc eased isk o leukemia.7,8 Fu he mo e, he e is
easonably consis en e idence o a sligh ly inc eased isk
associa ed wi h la ge bi h weigh ela i e o ges a ional
ime.9A highe isk has also been sugges ed o olde
pa en al age, deli e y by Cesa ean sec ion, and pa e nal
smoking.10–13 Howe e , dayca e a endance, alle gic dis-
eases, ma e nal olic acid supplemen a ion be o e bi h,
and ea ly immune s imula ion ha e been sugges ed o
educe he isk o leukemia.14–17
Al hough high doses o ionizing adia ion inc ease he
isk o childhood leukemia, he magni ude o any e ec
om low doses emains unce ain. Some s udies ha e
sugges ed inc eased isks associa ed wi h backg ound
adia ion and ollowing x- ay examina ions in u e o and
pos -na ally.18–22 Compu ed omog aphy (CT) imaging has
been used o almos ou decades and i s equency o
u iliza ion inc eased g ea ly du ing he 1980s-1990s. The
annual numbe o scans peaked a ound yea 2002; mo e
ecen ly CT scans ha e been pa ly eplaced by magne ic
esonance imaging in pedia ic imaging, pa ly because o
he isk o cance om ionizing adia ion.23 In 2015, 5,311
pedia ic CT scans we e pe o med in he Finnish popu-
la ion o 1,024,000 child en unde 17 yea s old, which is
a low a e compa ed o ha in many o he coun ies.23,24
Fou high-quali y s udies ha e in es iga ed he associa-
ion o pedia ic CT scans and childhood leukemia.25–28
The in e p e a ion o he indings mus include an e alu-
a ion o con ounding by indica ion, i.e. unde lying condi-
ions p edisposing child en o bo h CT scans and
leukemia.28–30 Ne e heless, he e idence is s ill limi ed
and he magni ude o he isk needs o be cha ac e ized
u he .
In his s udy, we examined he magni ude o he isk o
childhood leukemia a e pedia ic CT examina ions
using a na ionwide case-con ol design wi h e o s o
a oid e e se causa ion.
Me hods
We used a egis e -based, case-con ol s udy wi h indi idually
ma ched con ols. The key cha ac e is ics o he ma e ial ha e
been p esen ed p e iously.10 B ie ly, all cases o childhood
leukemia (M9800–M9948 in ICD-O-3) diagnosed in Finland du -
ing 1990–2011 (n=1,100) be o e he age o 15 yea s we e iden i-
ied om he Finnish Cance Regis y (Figu e 1). Th ee con ols
we e indi idually ma ched, by sex and yea o bi h, o each
case om he Popula ion Regis e Cen e . In all analyses, a 2-
A. Nikkilä e al.
1874 haema ologica | 2018; 103(11)
Figu e 1. Flow cha depic ing he selec ion o s udy
subjec s. The low o cases is ep esen ed on a whi e
backg ound and he con ols on a ligh g ay backg ound.
The necessa y exclusions a e shown in ed boxes.
Dashed lines ep esen he linking o he s udy subjec s
wi h he CT scans collec ed.
yea la ency pe iod was used, in pa o deal wi h e e se causa-
ion due o con ounding by indica ion.31 Also, mul iple p edis-
posing ac o s (Online Supplemen a y Table S1) we e accoun ed
o wi h ou pa ien egis e da a. The me hods a e desc ibed in
mo e de ail in he Online Supplemen a y Ma e ial.
We ob ained da a on all CT scans pe o med on pedia ic
pa ien s (<15 yea s) om all i e uni e si y hospi als and he i e
la ges cen al hospi als in Finland (Table 1, Figu e 2). The pe iod
o da a a ailabili y a ied be ween hospi als, because adiologi-
cal da abases wi h in o ma ion on each CT scan o indi idual
pa ien s we e in oduced a di e en imes. We es ima ed ha
he da a om he s udy hospi als co e ed 87% o all pedia ic
CT scans pe o med in Finland du ing 1975–2011 (see he Online
Supplemen a y Ma e ial o de ails). Fo each CT scan, we
Pedia ic CT scans and childhood leukemia
haema ologica | 2018; 103(11) 1875
Figu e 2. Flow cha linking he collec ed compu ed
omog aphy scans o he s udy subjec s The low
o he CT scans is ep esen ed on a whi e back-
g ound and he CT scans o di e en body pa s on
a ligh g ay backg ound. The necessa y exclusions
a e shown in ed boxes.
Table 1. The collec ion and a ailabili y o elec onically s o ed compu ed omog aphy scans.
Hospi al Ci y Da a a ailabili y Numbe o CT scans
Helsinki Uni e si y Hospi al Helsinki 1990–2011 31,825
Tampe e Uni e si y Hospi al Tampe e 1978–2011 9,236
Oulu Uni e si y Hospi al Oulu 1993–2011 7,513
Tu ku Uni e si y Hospi al Tu ku 1996–2011 7,360
Kuopio Uni e si y Hospi al Kuopio 1996–2011 5,408
Cen al Finland Cen al Hospi al Jy äskylä 2002–2011 2,571
Sa akun a Cen al Hospi al Po i 1995–2011 1,948
Seinäjoki Cen al Hospi al Seinäjoki 1999–2011 1,759
Päijänne Ta as ia Cen al Hospi al Lah i 2000–2011 1,597
No h Ka elia Cen al Hospi al Joensuu 1993–2011 1,191
TOTAL 80,783
All Finnish uni e si y hospi als a e lis ed i s ollowed by he i e chosen cen al hospi als.
ob ained he pa ame e s used o dose assessmen including
yea , body pa , use o con as medium and he numbe o
sequences. Manu ac u e s and models o CT scanne s in each
hospi al we e acqui ed om he Radia ion and Nuclea Sa e y
Au ho i y (STUK). Fo dose calcula ions, we assumed in he
main analysis ha each CT scan was pe o med using he la es
CT scanne a ailable a he hospi al.
Da a on a o al o 80,783 pedia ic CT scans we e ob ained and
o hose, 49 CT scans we e pe o med on he s udy subjec s,
excluding he 2-yea la ency pe iod (Table 1). Hal (n=25) we e
head scans, and 19 we e lung scans. O he CT scans, 36 we e pe -
o med on 15 (1.4%) cases and 13 scans on 10 (0.3%) con ols.
The CT scan pa ame e s we e ob ained based on expe opin-
ion o an expe ienced hospi al physicis (Online Supplemen a y
Table S2). The doses we e es ima ed using he NCICT so wa e
( 1.2).32 Age- and sex-speci ic pedia ic so wa e phan oms ( o
neona es, and child en aged 1, 5, 10, and 15 yea s) we e used.
The inpu o dose calcula ion also included he scanne manu-
ac u e and model. I da a we e a ailable only on he manu ac-
u e , a manu ac u e -speci ic a e age was used. I was assumed
ha a head o body il e was used, based on he a ge body
pa . The cumula i e abso bed ed bone ma ow (RBM) doses
we e ob ained as he sums o abso bed RBM doses om all CT
scans o each s udy subjec . The dose om a scan was mul i-
plied by 1.5 i con as medium was used, consis en wi h issue-
speci ic coe icien s sugges ed o o he issues.33 Al e na i e
dose es ima es we e ob ained based on alues epo ed in he li -
e a u e.34
We iden i ied subjec s wi h Down synd ome (40 cases and 2
con ols) om he Congeni al Mal o ma ion Regis e and Ca e
Regis e o Heal h Ca e, and hose wi h a p e ious malignancy
(2 cases) om he Finnish Cance Regis y. They we e excluded
o a oid con ounding by indica ion ( e e se causa ion). We also
collec ed in o ma ion on bi h weigh (la ge o ges a ional age)
and ma e nal smoking du ing p egnancy om he Medical Bi h
Regis e , as well as socioeconomic s a us and educa ion o he
pa en s om S a is ics Finland. Residen ial exposu e o back-
g ound gamma adia ion, including na u al e es ial adia ion
and Che nobyl allou , was es ima ed as desc ibed p e iously.8
Due o small equencies, exac condi ional logis ic eg ession
in SAS 9.4 was used o es ima ing odds a ios (OR), excess odds
a ios and hei con idence in e als (CI).35 S a is ical powe cal-
cula ions indica ed ha he sample size is su icien o de ec ing
a linea dose- esponse wi h an OR o 1.05 o g ea e pe 1 mGy
inc ease in cumula i e RBM dose wi h a s a is ical powe o
80% using asymp o ic condi ional logis ic eg ession.36
The e hical commi ee o Pi kanmaa Hospi al dis ic
e iewed he s udy p o ocol ( acking numbe R14074) and, in
acco dance wi h Finnish egula ions, no in o med consen was
equi ed o his egis e -based s udy. In addi ion, each hospi al
app o ed ou s udy p o ocol be o e deli e ing he da a on CT
scans. We ob ained pe mission o use da a om he Finnish
Cance Regis y, he Medical Bi h Regis e , Ca e Regis e o
Heal h Ca e and Congeni al Mal o ma ion Regis e om he
Na ional Ins i u e o Heal h and Wel a e (1774/5.05.00/2014), as
well as census da a on socioeconomic s a us om S a is ics
Finland (TK-52-306-16).
Resul s
In ou na ionwide egis e -based s udy, a e excluding
cases wi h an inco ec pe sonal iden i ica ion numbe o
p ohibi ion o use hei da a, we iden i ied 1,093 cases o
childhood leukemia diagnosed in 1990-2011. Mos o he
cases we e acu e lymphoblas ic leukemia (81.1%) o
acu e myeloid leukemia (13.0%). The median age a diag-
nosis among cases was 4.52 yea s (in e qua ile ange,
IQR 2.72 – 8.23). O he cases and con ols, 52% we e
male (Table 2). The c i e ia o la ge o ges a ional age
we e me by 121 (13.3%) o he cases and 275 (9.9%) o
he con ols.
A. Nikkilä e al.
1876 haema ologica | 2018; 103(11)
Table 2. The cha ac e is ics o cases and con ols be o e any exclusions.
Cases (n=1,093) Con ols (n=3,279) P
Gende
Female 48.0% (525) 48.0% (1575)
Male 52.0% (568) 52.0% (1704)
La ge o ges a ional age
No 86.7% (788) 90.1% (2493)
Ys 13.3% (121) 9.9% (275) 0.001
Missing 184 511
Mo he ’s smoked du ing
p egnancy
No 83.1% (742) 84.5% (2296)
Yes 16.9% (151) 15.5% (420) 0.096
Missing 200 563
Down synd ome
No 96.3% (1053) 99.9% (3277)
Yes 3.7% (40) 0.1% (2) <0.001
Pa en s’ educa ion
Mo he
Uppe seconda y 48.5% (530) 50.6% (1659) e
Bachelo ’s deg ee 22.3% (244) 23.1% (756) 0.869
Mas e ’s o doc o ’s deg ee 10.2% (112) 9.8% (321) 0.406
Missing 18.9% (207) 16.6% (543)
Fa he
Uppe seconda y 52.0% (568) 51.4% (1685) e
Bachelo ’s deg ee 15.2% (166) 16.2% (532) 0.423
Mas e ’s o doc o ’s deg ee 10.0% (110) 10.2% (334) 0.880
Missing 22.8% (249) 22.2% (728)
Pa en s’ socioeconomic s a us
Mo he
Sel -employed 7.7% (84) 8.3% (273) e
Uppe le el employee 16.1% (176) 15.7% (514) 0.477
Lowe le el employee 34.8% (380) 34.5% (1130) 0.521
Manual wo ke 21.4% (231) 20.6% (674) 0.490
O he 18.2% (199) 20.3% (664) 0.859
Missing 2.1% (23) 0.7% (24)
Fa he
Sel -employed 13.9% (152) 12.0% (395) e
Uppe le el employee 17.6% (192) 18.2% (596) 0.178
Lowe le el employee 18.3% (197) 17.9% (587) 0.273
Manual wo ke 34.0% (372) 35.0% (1148) 0.170
O he 12.4% (135) 14.3% (469) 0.036
Missing 4.1% (45) 2.6% (84)
Age a leukemia diagnosis, yea s
0 – 2 14.3% (156)
2 – 7 55.5% (605)
7 – 15 33.4% (332)
Leukemia ype
P e-B-ALL 75.6% (826)
P e-T-ALL 5.9% (64)
Unclassi ied ALL 1.8% (20)
Acu e myeloid leukemia 13.6% (149)
O he 3.1% (34)
The epo ed P- alues a e om an uni a ia e condi ional logis ic eg ession model. The non-
bina y a iables we e ea ed as ac o s and he e e ence ca ego ies a e ma ked wi h “ e ”. ALL:
acu e lymphoblas ic leukemia.
A e exclusions, eigh cases (0.7%) and nine con ols
(0.3%) had unde gone a leas one CT scan. The median
RBM dose was 10.1 mGy (IQR 4.79 – 13.6) o he
exposed cases and 6.29 mGy (IQR 5.69 – 7.14) o he
exposed con ols (Figu e 3). The co esponding li e a u e-
based alues we e 26.5 mGy and 17.6 mGy. The RBM
doses calcula ed wi h NCICT om ho acic CT scans a -
ied be ween 1.8 and 6.8 mGy (median 4.0 mGy) and sim-
ila ly, he doses o head CT scans a ied be ween 1.6 and
10.7 mGy (median 6.6 mGy).
The OR o any e sus no CT was 2.82 (95% CI: 1.05 –
7.56). Fo wo o mo e pedia ic CT scans, he OR was
5.22 (95% CI: 0.89 – 69.9). Fo any head CT scans, an OR
o 4.00 (95% CI: 1.39 – 11.5) was ob ained.
The o e all excess OR o childhood leukemia was 0.13
(95% CI: 0.02 – 0.26) pe mGy o abso bed RBM dose cal-
cula ed wi h he NCICT so wa e (Table 3). Using he
cumula i e RBM dose es ima es om he li e a u e, an
excess OR o 0.05 (95% CI: 0.01 – 0.10) pe mGy was
ob ained. In an analysis by dose e ile calcula ed wi h
NCICT, he excess OR ela i e o ze o dose we e 1.26 (95%
CI: -0.50 – 10.1) o he i s g oup, 0.09 (95% CI: -0.89 –
10.5) o he second, and 5.00 (95% CI: 0.10 – 31.7) o he
las (Figu e 4).
Fo he mos common sub ype, p ecu so B-cell acu e
lymphoblas ic leukemia, he excess OR pe mGy was
0.14 (95% CI: 0.02 – 0.29) using es ima es om NCICT
and 0.06 (0.01 – 0.11) o li e a u e-based es ima es. The
excess OR o any e sus no CT scans was 2.25 (95% CI:
0.08 – 8.75) o acu e lymphoblas ic leukemia and 2.88
(95% CI: 0.22 – 11.4) o p ecu so B-cell acu e lym-
phoblas ic leukemia. In he analysis by age a
diagnosis/ e e ence da e, he excess OR o any e sus no
CT scans was 3.50 (95% CI: -0.25 – 25.9) o child en
aged 2 – <7 yea s and 1.27 (95% CI: -0.32 – 6.54) o
hose aged 7 – <15 yea s.
Co a ia e (con ounde ) adjus men s (la ge o ges a-
ional age, ma e nal smoking du ing p egnancy, pa en al
educa ion and pa en al socioeconomic s a us) did no
al e he OR o CT exposu e by mo e han 0.05 uni s,
wi h he excep ion o ma e nal smoking, which inc eased
he OR ela ed o he numbe o pedia ic CT scans (0 e -
sus 1 o mo e) (app oxima ely 0.10 uni s). Ne e heless,
we p e e ed he unadjus ed model, as missing da a on
ma e nal smoking esul ed in wide con idence in e als
o he main a iables.
The OR we e highe when he subjec s wi h Down
synd ome we e no excluded ( o 1 o mo e CT scans
OR=5.21, 95% CI: 2.19 – 12.4 and o cumula i e RBM
dose excess OR=0.19 pe mGy, 95% CI: 0.07 – 0.32). No
e idence o a di e en e ec o he RBM doses on
leukemia isk o subjec s wi h o wi hou Down syn-
Figu e 3. His og ams o (A) he ages o he subjec s a he ime o compu ed
omog aphy scan and (B) he cumula i e doses ecei ed by he subjec s, calcu-
la ed wi h NCICT.
A
B
Table 3. The equencies o compu ed omog aphy scans o subjec s >2
yea s old a he e e ence da e and odds a ios calcula ed wi h exac condi-
ional logis ic eg ession.
Cases Con ols OR (95% CI)
911 2730
CT scans
0 903 2721
1 4 7 1.85 (0.39, 7.36)
2 o mo e 4 2 6.22 (0.89, 68.9)
by ype (1 o mo e)
ALL 7 7 3.25 (1.08, 9.75)
p e-B-ALL 7 6 3.88 (1.22, 12.4)
by age-g oup (1 o mo e)
2 - <7 yea s 3 2 4.50 (0.75, 26.9)
7 - <15 yea s 5 7 2.27 (0.68, 7.54)
by dose index (NCICT/li e a u e)
low, 4.79a/11.6bmGy 3 4 2.26 (0.50, 10.1)
medium, 6.72a/19. 8bmGy 1 3 1.09 (0.11, 10.5)
high, 13.8a/33.2bmGy 4 2 6.00 (1.10, 32.8)
pe 1 mGy (NCICT)
TOTAL 1.13 (1.02, 1.26)
p e-B-ALL 1.14 (1.02, 1.29)
pe 1 mGy (li e a u e)
TOTAL 1.05 (1.01, 1.10)
p e-B-ALL 1.06 (1.01, 1.11)
The e e ence g oup o all calcula ed odds a io (OR) is ze o CT scans o ca ego ical a i-
ables. S udy subjec s wi h Down synd ome o cance diagnoses we e excluded. All epo ed
OR a e om an unadjus ed model. The median doses o dose-index classes calcula ed wi h
NCICT a e ma ked wi h a. The espec i e class medians based on li e a u e a e ma ked wi h b.
ALL: acu e lymphoblas ic leukemia; p e-B-ALL: p ecu so B-cell acu e lymphoblas ic
leukemia.
Pedia ic CT scans and childhood leukemia
haema ologica | 2018; 103(11) 1877

d ome was ound o sugges e ec modi ica ion (in e ac-
ion P=0.99).
When he oldes possible CT scanne (a maximum, 10
yea s old) a he hospi al was used in dose es ima ion
ins ead o he mos mode n CT scanne , he median
cumula i e RBM dose o cases was 9.71 mGy (IQR 7.09
– 18.7) and o con ols 7.14 mGy (IQR 5.71 – 12.6), wi h
an excess OR o 0.11 (95% CI: 0.02 – 0.22) pe mGy.
When he cumula i e RBM dose om e es ial
gamma adia ion and Che nobyl allou was included in
he model, he OR o cumula i e RBM dose om pedi-
a ic CT scans emained unchanged. The median cumula-
i e dose om esiden ial gamma adia ion was 1.96 mS
o cases and 1.90 mS o con ols.
The dis ibu ions o ci ies o he las add esses o cases
and con ols we e analyzed o e alua e whe he cases
and con ols belonged o ca chmen popula ions o di e -
en hospi als, which migh ha e caused di e en ial mis-
classi ica ion due o con as ing da a a ailabili y. No di -
e ence in he dis ibu ions was no ed (chi-squa ed es ,
P=0.30). The age and CT scan yea s o he subjec s a e
epo ed in Online Supplemen a y Table S3.
Discussion
We es ima ed he impac o adia ion exposu e om
pedia ic CT scans on isk o childhood leukemia in a
na ionwide egis e -based case-con ol s udy in Finland.
O e all, a s a is ically signi ican inc ease in isk pe mGy
o RBM abso bed dose was ound. The cen al es ima e is
la ge han in p e ious s udies, bu he con idence in e -
als o e lap wi h ea lie esul s, and he e ec size is
compa ible wi h ex apola ion om high-dose s udies.
The highe main poin es ima e is likely in luenced by
andom e o , as he dose es ima es we e imp ecise due
o lack o de ail in dosime ic da a, including pa ame e
alues used o he scanne . I is also possible ha he yp-
ical alues based on expe opinion a e ep esen a i e o
cu en p ocedu es, bu may unde es ima e doses om
olde examina ions, which could in la e he isk es ima es
pe uni dose. Howe e , ou si e-speci ic dose es ima es
calcula ed wi h NCICT we e qui e compa able wi h
hose epo ed in he B i ish s udy.25 We minimized he
po en ial o sys ema ic e o by adjus ing o se e al con-
ounde s and used consis en p ocedu es o he cases and
con ols. The isk es ima es we e sligh ly highe o p e-
cu so B-cell acu e lymphoblas ic leukemia han o o he
leukemias, bu he di e ence was no s a is ically signi i-
can .
Two la ge s udies ha e been published on he subjec
p io o ou s. The coho s udies om he Uni ed
Kingdom and Aus alia epo ed a signi ican isk o child-
hood leukemia associa ed wi h RBM dose om pedia ic
CT scans.25,26 Pea ce e al. ound an excess ela i e isk o
0.04 pe mGy and Ma hews e al. epo ed a ela i e isk
o 1.2 o one o mo e CT scans wi h an excess ela i e
isk o 0.04 pe mGy. The Aus alian coho had 211
exposed leukemia cases and he UK s udy 74. A smalle
Ge man coho s udy epo ed an inc eased leukemia
incidence ollowing wo o mo e CT scans, bu a non-sig-
ni ican dose- esponse based on 12 exposed cases.27 Based
on he Li e Span S udy in Japan, he ex apola ed excess
ela i e isk o childhood exposu e would be app oxi-
ma ely 0.05 pe mGy.37
O he majo sou ces o ionizing adia ion we e aken
in o conside a ion by including cumula i e RBM doses
om e es ial gamma adia ion and Che nobyl allou ,
and his did no a ec he esul s. In ou da a, he a e age
cumula i e RBM dose om CT o he con ols was only
0.002 mGy, which is app oxima ely 0.1% o he a e age
annual RBM dose in Finland.38 We accoun ed o medical
use o adia ion, o which omog aphy scans make he
la ges con ibu ion, and e es ial gamma adia ion,
which accoun s o nea ly wo- hi ds o a e age annual
adia ion o he RBM in Finland.23,39 In addi ion, he e is
li le e idence o assume ha o he sou ces o ionizing
adia ion, such as cosmic adia ion o in e nal exposu e o
na u al adioiso opes, would dis ibu e unequally among
he cases and con ols.
The coe icien 1.5 o inc emen al dose due o CT
imaging wi h con as medium was chosen p agma ically
based on he coe icien s o o he body pa s, as he
e ec s on RBM dose we e no epo ed sepa a ely.33
Based on limi ed popula ion s a is ics a ailable om
he Radia ion and Nuclea Sa e y Au ho i y,23 oughly 30
CT scans we e expec ed o he con ols. Howe e , only
13 scans we e eco ded among hem. This migh pa ly
e lec incomple e a ailabili y o da a, bu he es ima e o
he expec ed numbe s is highly unce ain because o lack
o da a on pedia ic CT scans p io o 2008. I is also
wo h no ing ha pedia ic CT scans a e pe o med less
equen ly in Finland han in se e al o he coun ies.24
Ou ma e ial consis s o a comp ehensi e se o child-
hood leukemia cases and ep esen a i e con ols, which
should elimina e selec ion bias by i ue o a egis e -
based app oach, which equi ed no consen o in o ma-
ion om he s udy subjec s o hei amilies. The s udy
pe iod co e s he yea s in which he use o pedia ic CT
scans was mos equen , as he annual numbe o pedi-
a ic CT scans has been dec easing in Finland since he
yea 2000.23 The da a on CT scans we e ob ained om
A. Nikkilä e al.
1878 haema ologica | 2018; 103(11)
Figu e 4. Dose- esponse cu e o cumula i e ed bone ma ow dose om pedi-
a ic compu ed omog aphy scans and childhood leukemia. The poin es i-
ma es wi h 95% con idence in e als a e o he h ee dose index le els and he
i ed cu e is o he cumula i e RBM dose calcula ed wi h NCICT. The shaded
a ea ep esen s he 95% con idence in e al o he con inuous dose- esponse.
The e ical axis is on a bina y loga i hm scale.
hospi al da abases o a oid ecall bias, and also included
he scanne model and use o con as medium. As in
o he s udies, he mos common single CT scan in ou
analysis was a head scan.23
Radia ion doses o RBM om he CT scans we e calcu-
la ed using he bes a ailable me hods, employing NCICT
so wa e, wi h age- and sex-speci ic phan oms and aking
in o accoun he scanne model. The scanning pa ame e s
en e ed in o he so wa e we e based on he se ings and
p ocedu es commonly used in Finland, al hough da a we e
no a ailable o each scan. We also e alua ed he e ec s o
choosing he mos mode n CT scanne a each imaging si e
and he OR showed obus beha io .
We also had da a on se e al impo an isk ac o s includ-
ing Down synd ome, pa en al socioeconomic s a us, la ge
o ges a ional age and ma e nal smoking. We we e able o
inco po a e da a on cance p edisposing ac o s, which
ha e been shown o be o impo ance ecen ly.28,30 Inclusion
o cases wi h Down synd ome would ha e inc eased he
isk es ima es, possibly because Down synd ome is associ-
a ed wi h inc eased isks o bo h leukemia and in ec ions.4,40
We also explo ed he join e ec o Down synd ome and
cumula i e RBM dose and ound no in e ac ion. Subg oup
analyses o explo a o y na u e we e ca ied ou by sub ype
o childhood leukemia and age a diagnosis, al hough
hese we e unde powe ed.
Ou s udy has some sho comings. We we e able o
ob ain da a om all en hospi als only a e 2002, hus
exposu e assessmen is no uni o mly comple e o sub-
jec s bo n p io o ha yea . Only a mino imp o emen
in s a is ical powe would ha e been eached by collec -
ing pedia ic CT scans om he es o he imaging cen-
e s in Finland. In addi ion, he e is no eason o assume
ha he missed CT scans would ha e been unequally dis-
ibu ed o he cases and con ols, i.e. esul in di e en-
ial misclassi ica ion. Fo dose es ima ion, comple e in o -
ma ion on he scanning pa ame e s is included in he
mode n pic u e a chi ing sys ems, bu was no a ailable
be o e he yea 2000. Use o pa ame e s o each indi id-
ual scan would ha e p o ided mo e accu a e dose es i-
ma es. The unexpec edly lowe median dose o cases o
olde scanne s ound in ou sensi i i y analysis may be
due o andom e o . The numbe o di e en CT scan-
ne s in ou analysis was limi ed and hus he es ima es o
a e age dose we e imp ecise.
Ou esul s suppo he no ion ha e en small doses o
adia ion om pedia ic CT scans p oduce a small, bu
de ec able inc ease in leukemia isk. In he subg oup
analyses, we obse ed no subs an ial di e ences by age
o leukemia sub ype, al hough sligh ly highe isks we e
ound o p ecu so B-cell acu e lymphoblas ic leukemia.
Acknowledgmen s
The au ho s would like o hank Isabelle Thie y-Che
(IARC), Hannele Niini ii a (Tu ku Uni e si y Hospi al), Juha
Suu a i (STUK) and Rebecca Smi h-Bindman (Uni e si y o
Cali o nia, San F ancisco) o hei aluable sugges ions and
commen s ega ding modeling o CT scan doses wi h he da a
a ailable in Finland. We a e also g a e ul o D Choonsik Lee
(Na ional Cance Ins i u e) o his insigh ul commen s ela ed o
modeling con as media and o p o iding us wi h his s a e-o -
he-a dose calcula ion so wa e (NCICT) and Anni iikka
Ran ala (STUK) o collec ing he da a on CT scanne s used in
Finland. Päi i Laa ne’s (Tampe e Uni e si y Hospi al) c ucial
inpu ega ding he scanning pa ame e s enabled us o use
NCICT so wa e. Funding o he s udy was ob ained om he
Finnish Founda ion o Pedia ic Resea ch, Vä e Founda ion o
Pedia ic Cance Resea ch and Compe i i e S a e Resea ch
Financing o he Expe Responsibili y a ea o Tampe e
Uni e si y Hospi al (9T030 and 9U030).
Pedia ic CT scans and childhood leukemia
haema ologica | 2018; 103(11) 1879
Re e ences
1. Eden T. Ae iology o childhood leukaemia.
Cance T ea Re . 2010;36(4):286–297.
2. Cance Incidence, Mo ali y, P e alence and
Su i al in he No dic Coun ies, Ve sion 8.0
(20.12.2017).
3. Hsu W-L, P es on DL, Soda M, e al. The
incidence o leukemia, lymphoma and mul-
iple myeloma among a omic bomb su -
i o s: 1950-2001. Radia Res. 2013;179(3):
361–382.
4. Hasle H. Pa e n o malignan diso de s in
indi iduals wi h Down’s synd ome. Lance
Oncol. 2001;2(7):429–436.
5. Haddy N, Cécile Le Deley M, Samand A, e
al. Role o adio he apy and chemo he apy
in he isk o seconda y leukaemia a e a
solid umou in childhood. Eu J Cance .
2006;42(16):2757–2764.
6. Kamiya K, Ozasa K, Akiba S, e al. Long-
e m e ec s o adia ion exposu e on heal h.
Lance . 2015;386(9992):469–478.
7. Wiemels JL, Walsh KM, de Smi h AJ, e al.
GWAS in childhood acu e lymphoblas ic
leukemia e eals no el gene ic associa ions
a ch omosomes 17q12 and 8q24.21. Na
Commun. 2018;9(1):286.
8. Rippe ge T, Bielack SS, Bo kha d A, e al.
Childhood cance p edisposi ion syn-
d omes-A concise e iew and ecommenda-
ions by he Cance P edisposi ion Wo king
G oup o he Socie y o Pedia ic Oncology
and Hema ology. Am J Med Gene A.
2017;173(4):1017–1037.
9. Milne E, G eenop KR, Me aye C, e al. Fe al
g ow h and childhood acu e lymphoblas ic
leukemia: indings om he Childhood
Leukemia In e na ional Conso ium. In J
Cance . 2013;133(12):2968-2979.
10. Nikkilä A, E me S, A ela H, e al.
Backg ound adia ion and childhood
leukemia: a na ionwide egis e -based case-
con ol s udy. In J Cance . 2016;139(9):
1975–1982.
11. Ma co e EL, Thomopoulos TP, In an e-
Ri a d C, e al. Caesa ean deli e y and isk
o childhood leukaemia: a pooled analysis
om he Childhood Leukemia In e na ional
Conso ium (CLIC). Lance Haema ol.
2016;3(4):e176–e185.
12. Kendall GM, Li le MP, Wake o d R, e al. A
eco d-based case-con ol s udy o na u al
backg ound adia ion and he incidence o
childhood leukaemia and o he cance s in
G ea B i ain du ing 1980 – 2006. Leukemia.
2013;27(1):3–9.
13. Me aye C, Pe idou E, A angu é JMM, e al.
Pa en al obacco smoking and acu e myeloid
leukemia: he Childhood Leukemia
In e na ional Conso ium. Am J Epidemiol.
2016;184(4):261–273.
14. Me aye C, Milne E, Docke y JD, e al.
Ma e nal supplemen a ion wi h olic acid
and o he i amins and isk o leukemia in
o sp ing: a Childhood Leukemia
In e na ional Conso ium s udy.
Epidemiology. 2014;25(6):811–822.
15. Rudan J, Ligh oo T, U ayama KY, e al.
Childhood acu e lymphoblas ic leukemia
and indica o s o ea ly immune s imula ion:
a Childhood Leukemia In e na ional
Conso ium s udy. Am J Epidemiol. 2015;
181(8):549–562.
16. Ma X, Bu le PA, Sel in S, e al. Dayca e
a endance and isk o childhood acu e lym-
phoblas ic leukaemia. B J Cance . 2002;
86(9):1419–1424.
17. Linabe y AM, Ju ek AM, Du al S, Ross JA.
The associa ion be ween a opy and child-
hood/adolescen leukemia: a me a-analysis.
Am J Epidemiol. 2010;171(7):749–764.
18. Shu XO, Po e JD, Line MS, e al. Diagnos ic
X- ays and ul asound exposu e and isk o
childhood acu e lymphoblas ic leukemia by
immunopheno ype. Cance Epidemiol
Bioma ke s P e . 2002;11(2):177–185.
19. Meine R, Kale sch U, Kaa sch P, Schüz J,
Michaelis J. Associa ions be ween child-
hood cance and ionizing adia ion: esul s
o a popula ion-based case-con ol s udy in
Ge many. Cance Epidemiol Bioma ke s
P e . 1999;8(9):793–799.
20. In an e-Ri a d C, Ma honne G, Sinne D.
Risk o childhood leukemia associa ed wi h
diagnos ic i adia ion and polymo phisms in
DNA epai genes. En i on Heal h Pe spec .
2000;108(6):495.
21. Ba ley K, Me aye C, Sel in S, Duco e J,
Bu le P. Diagnos ic X- ays and isk o child-
hood leukaemia. In J Epidemiol. 2010;39(6):
1628–1637.
22. Ha b on RW, Chapple C-L, O’Sulli an JJ, e
al. Cance incidence among child en and
young adul s who ha e unde gone x- ay
guided ca diac ca he e iza ion p ocedu es.
Eu J Epidemiol. 2018;33(4):393-401.
23. Suu a i J. Radiologis en u kimus en ja
oimenpi eiden mää ä uonna 2015. STUK;
2015.
24. UNSCEAR. Sou ces, E ec s and Risks o
Ionizing Radia ion, UNSCEAR 2017 Repo .
Uni ed Na ions.
25. Pea ce MS, Salo i JA, Li le MP, e al.
Radia ion exposu e om CT scans in child-
hood and subsequen isk o leukaemia and
b ain umou s: a e ospec i e coho s udy.
Lance . 2012;380(9840):499–505.
26. Ma hews JD, Fo sy he AV, B ady Z, e al.
Cance isk in 680,000 people exposed o
compu ed omog aphy scans in childhood
o adolescence: da a linkage s udy o 11 mil-
lion Aus alians. BMJ. 2013;346: 2360.
27. K ille L, D ege S, Schindel R, e al. Risk o
cance incidence be o e he age o 15 yea s
a e exposu e o ionising adia ion om
compu ed omog aphy: esul s om a
Ge man coho s udy. Radia En i on
Biophys. 2015;54(1):1–12.
28. Jou ny N, Rehel J-L, Ducou Le Poin e H, e
al. A e he s udies on cance isk om CT
scans biased by indica ion? Elemen s o
answe om a la ge-scale coho s udy in
F ance. B J Cance . 2015;112(1):185–193.
29. Be ing on De Gonzalez A, Salo i JA,
Mchugh K, e al. Rela ionship be ween pae-
dia ic CT scans and subsequen isk o
leukaemia and b ain umou s: assessmen o
he impac o unde lying condi ions. B J
Cance . 2016;114(4):388-394.
30. Jou ny N, Roué T, Ca dis E, e al. Childhood
CT scans and cance isk: impac o p edis-
posing ac o s o cance on he isk es i-
ma es. J Radiol P o . 2016;36(1):N1–N7.
31. ANNEx BScien i. UNSCEAR 2013 Repo .
32. Lee C, Kim KP, Bolch WE, Mo oz BE, Folio
L. NCICT: a compu a ional solu ion o es i-
ma e o gan doses o pedia ic and adul
pa ien s unde going CT scans. J Radiol P o .
2015;35(4):891-909.
33. Ama o E, Salamone I, Naso S, Bo a i A,
Gae a M, Blandino A. Can con as media
inc ease o gan doses in CT examina ions? A
clinical s udy. Am J Roen genol. 2013;200(6):
1288–1293.
34. Kim KP, Be ing on de González A, Pea ce
MS, e al. De elopmen o a da abase o
o gan doses o paedia ic and young adul
CT scans in he Uni ed Kingdom. Radia
P o Dosime y. 2012;150(4):415–426.
35. Meh a CR, Pa el NR. Exac logis ic eg es-
sion: heo y and examples. S a Med.
1995;14(19):2143–2160.
36. Lachin JM. Sample size e alua ion o a mul-
iply ma ched case-con ol s udy using he
sco e es om a condi ional logis ic (dis-
c e e Cox PH) eg ession model. S a Med.
2008;(27):2509–2523.
37. Wake o d R, Li le MP, Kendall GM. Risk o
childhood leukemia a e low-le el exposu e
o ionizing adia ion. Expe Re Hema ol.
2010;3(3):251–254.
38. Muikku M, A ela H, Jä inen H, e al.
Annoskakku 2004 - Suomalais en
keskimää äinen e ek ii inen annos. 2005.
39. Muikku M, Bly R, Lah inen J, e al.
Suomalais en keskimää äinen e ek ii inen
annos Suomalais en keskimää äinen e ek i-
i inen annos. 2014.
40. Ram G, Chinen J. In ec ions and immuno -
de iciency in Down synd ome. Clin Exp
Immunol. 2011;164(1):9–16.
A. Nikkilä e al.
1880 haema ologica | 2018; 103(11)