Association of Paternal Diagnostic X-Ray Exposure with ... · leukemia, higher risks being linked to exposures closer to conception. X-ray related leukemia risk varied with exposure

Vol. 3, 645- 653, December 1994 Cancer Epidemiology, Biomarkers & Prevention 645

Association of Paternal Diagnostic X-Ray Exposure withRisk of Infant Leukemia’

Xiao-Ou Shu,2 Gregory H. Reaman, Beatrice Lampkin,Harland N. Sather, Thomas W. Pendergrass, andLeslie L. Robison for the investigators of the ChildrensCancer Group3Divisions of Pediatric Epidemiology/Clinical Research and Oncology,University of Minnesota, Minneapolis, Minnesota ]X-O. S., L. L. R.];

Department of Pediatric Hematology-Oncology, Children’s NationalMedical Center, Washington, DC 1G. H. RI; Department of PediatricHematology-Oncology, Children’s Hospital of Cincinnati, Cincinnati,Ohio (B. LI; Department of Preventive Medicine, University of SouthernCalifornia School of Medicine, Los Angeles, California ]H. N. S.]; andDepartment of Pediatric Hematology-Oncology, Children’s Hospital andMedical Center, Seattle, Washington IT. W. P.]

Abstract

Whether low level radiation exposure before conceptionincreases the risk of leukemia in offspring has beenmuch debated. No study has specifically evaluated theeffed of parental preconception diagnostic X-rayexposure in the development of leukemia among infants.Mothers of 302 infant leukemia cases (diagnosed at �1 8months of age) and 558 individually matched regionalcontrols, and fathers of 250 cases and 361 controls,were independently interviewed to obtain informationon X-ray exposures. Paternal preconception X-rayexposure was associated with an increased risk of infantleukemia, higher risks being linked to exposures closerto conception. X-ray related leukemia risk varied withexposure site and histopathological type, the highest riskbeing for acute lymphocytic leukemia related to two ormore X-rays of the lower gastrointestinal (GI) trad andlower abdomen (odds ratio, 3.78; 95% confidenceinterval, 1 .49-9.64). A positive association was observedbetween acute lymphocytic leukemia and number ofpaternal X-rays of the lower GI and lower abdomen(trend test, P < 0.01 ), upper GI (P = 0.04), and chest(P = 0.08). Exposures of head and neck and limbs wereunrelated to risk. The risk of acute myelogenousleukemia was unrelated to paternal X-ray exposure,except for a marginally significant association (trendtest, P = 0.07) for upper Gl X-rays. No consistentassociation between maternal X-ray exposure and infantleukemia was observed. The results of this study suggestthat paternal low level radiation exposure beforeconception is associated with an increased risk of infant

Received 6/28/94; revised 9/12/94; accepted 9/12/94.

1 Supported by National Cancer Institute Grant CA4279, the University of

Minnesota Children’s Cancer Research Fund, and the United States Depart-ment of Health and Human Services.2 To whom requests for reprints should be addressed, at Childrens Cancer

Group, P.O. Box 60012, Arcadia, CA 91 066-601 2.

3 Contributing Childrens Cancer Group investigators, institutions, and grantnumbers are given in the Appendix.

leukemia, although the nature of this as3ociation needsto be further evaluated.

IntrodudionThe finding of a 6-fold elevation in risk of childhood leu-

kemia in relation to paternal occupational exposure of morethan 100 mSv (low dose) of ionizing radiation during em-ployment at a nuclear plant in Sellafield, United Kingdom(1 ), has led to a debate over low dose radiation as a cause

of germinal mutation. Controversy exists because of theabsence of increased risk of leukemia in offspring of Japa-nese atomic bomb survivors (2) and in offspring of nuclear

workers in other settings (3, 4), and lack of statistical corn-patibility of the distribution of paternal preconceptionalradiation exposure dose and the clustering of childhoodleukemia occurring in Seascale (5). The association be-

tween paternal diagnostic X-ray exposure and risk of child-hood cancer has received little attention (6, 7).

If preconception radiation exposure is indeed relatedto the risk of leukemia in offspring, we might expect toobserve a stronger association among the youngest children(i.e., those diagnosed in the first 2 years of life). Age atdiagnosis of leukemia, however, has not been the focus of

this debate, although leukemia cases with paternal precon-ceptional radiation exposure tended to be younger thanthose without such exposure (8). None of the previous

studies had sufficient study power to investigate the asso-ciation of preconception radiation exposure and risk ofleukemia among very young children.

The Childrens Cancer Group recently conducted alarge case-control study of leukemia among children di-agnosed before 1 8 months of age. Infants provide a

unique opportunity to study genetic and prenatal riskfactors of leukemia, with minimum influence of postnatal

exposures. We present here data on the association be-tween parental diagnostic radiation exposure and infantleukemia risk.

Subjeds and MethodsCases were identified through the registration files of the

CCG,4 a cooperative clinical trials group with approxi-mately 1 00 members and affiliated institutions in the United

States and Canada. To be eligible for this study, cases hadto be newly diagnosed with leukemia between January 1,1 983 and December 31 , 1 988 and be 1 8 months of age or

younger. Additional eligibility criteria were that there had to

be a telephone in the residence of the patient and that the

4 The abbreviations used are: CCG, Childrens Cancer Group; GI, gastroin-testinal; OR, odds ratio; Cl, confidence interval; ALL, acute lymphocytic

leukemia; AML, acute myelogenous leukemia.

on February 27, 2021. © 1994 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

http://cebp.aacrjournals.org/

646 Paternal X-ray Exposure and Infant Leukemia

biological mother of the patient had to be available for

interview and speak English. A total of 492 cases were

identified during the study period, and 382 of them met the

eligibility criteria. Telephone interviews with mothers were

completed for 302 cases (79.1%). The remaining 80 cases

were not interviewed due to physician refusal (n = 19),

parental refusal (n = 30), inability to locate the parents (n =

18), and other miscellaneous reasons (n = 13).

Controls were randomly selected, using a previously

described random digit dialing procedure (9) and individ-

ually matched to cases in a 2:1 ratio on year of birth,

telephone area code, and exchange. As with the cases,

there had to be a telephone in the residence of the control

and the biological mother had to be available for interview

and speak English. A total of 743 eligible controls were

identified. Of these, the mothers of 558 (75.1%) children

were interviewed, which resulted in 50 sets of 1 :1 , 249 sets

of 1 :2, 2 sets of 1 :3, and 1 set of 1 :4 case-control matching

sets. The major reason for nonparticipation of controls was

parental refusal (18.6%).

Data were collected by independent telephone inter-

views with mothers (and fathers whenever available) of

cases and controls using structured questionnaires. The

questionnaire for the mother covered demographics,

events, and exposures prior to and during the index preg-

nancy and birth, lifetime occupation history, reproductive

history, and family medical history. The questionnaire forthe father focused on medication use, X-ray exposure, and

personal habits prior to the index pregnancy, lifetime oc-cupation history, and family medical history. The question-

naires for the fathers were completed for 280 of the 382

eligible cases (73%) and 51 1 of the 743 eligible controls

(69%). Of these totals, direct interviews with fathers were

obtained for 89% of cases and 71 % of controls. The re-

maining interviews were completed by mothers as surro-

gates for the fathers.

Detailed information on parental preconception diag-

nostic X-ray exposure and in utero radiation exposure was

collected. The parents were asked whether they had X-ray

exposure within a month, a year, or more than a year prior

to the index pregnancy. If X-ray exposure had occurred,information was collected on anatomic sites, major reason

for the X-rays, and the cumulative number of exposures at

each specific site. !n utero exposures were collected in the

mother’s interview, including questions on the numberof X-ray exposures at different sites and ultrasound exami-

nations during pregnancy. Self-reported information onoccupational exposure to radioactive materials also was

obtained.

In the data analyses, X-ray exposures of some anatomic

sites with low exposure rates were combined into catego-ries comprising broader anatomic regions. For example,

lower Gl barium enema, kidney i.v. pyelogram, and X-ray

of the hip were classified as lower CI tract and abdomen

exposure; head and neck X-ray included dental X-rays, aswell as X-rays of other parts of the head and neck. ORs, as

approximations of relative risk, were used to measure the

association between X-ray exposure and risk of infant leu-

kemia. Conditional logistic regression was employed in

data analyses to obtain ORs and 95% CIs, adjusting for

potential confounders (1 0). Tests for trend were performed

by treating levels of categorical variables as a continuousvariable in the logistic model.

Table 1 Risk of infant leukemia associated with demographic

characteristics

c�;:�s ControlsOR” 95% Cl

Maternal age

<25 34.4 27.8 1.0

25-29 36.4 37.1 0.76 0.54-1.08

30-34 21.5 26.7 0.62 0.41-0.94

35+ 7.6 8.4 0.73 0.40-1.25

Maternal education

<High school 13.9 8.4 1.0

High school 35.8 31 .1 0.66 0.40-1.09

>High school 50.3 60.5 0.45 0.28-0.74

Maternal smoking

Never 54.8 50.3 1.0

Ever 45.2 49.7 0.84 0.62-1.13

Maternal drinking

Never 9.0 14.3 1.0

Ever 91.0 85.7 1.78 1.09-2.91

Paternal age

<25 20.4 16.9 1.0

25-29 29.0 32.6 0.71 0.45-1.12

30-34 33.3 31.8 0.83 0.52-1.33

35+ 17.2 18.7 0.74 0.43-1.27

Paternal education

<High school 10.8 7.5 1.0

High school 33.0 30.0 0.80 0.44-1.48

>High school 56.3 62.5 0.60 0.33-1.09

Paternal smoking

Never 42.7 47.9 1.0

Ever 57.3 52.1 1.22 0.90-1.66

Paternal drinking

Never 3.9 8.0 1.0

Ever 96.1 92.0 2.07 1.04-4.12

Family income ($/yr)

�22,000 32.1 27.8 1.0

22,000-35,000 26.5 31.0 0.68 0.46-1.01

>35,000 23.8 29.9 0.68 0.45-1.03

Unknown 17.6 11.3 1.33 0.83-2.12

a Frequencies were obtained for all cases and controls pooled, ignoringmatching status. Subjects with missing values were excluded.b ORs were derived from conditional logistic regression model.

Results

Demographic Charaderistics. Of the 302 cases includedin this study, 203 were diagnosed with ALL and 88 withAML. The leukemia type was not clear for 1 1 patients. Of

these cases, 91 (30%) were younger than 7 months, 96(32%) were between 7 and 1 2 months, and 1 1 5 (38%) were

between 1 3 and 1 8 months old at diagnosis of leukemia.There were slightly more girls than boys (female/male =

1 .1 0) in cases but more boys in the control group (female/male = 0.78). Case mothers were significantly younger than

control mothers at the birth of the index child (mean ma-ternal age was 26.7 for cases and 27.5 for controls; P =

0.02), but a test for trend in risk of infant leukemia bymaternal age category was not significant. Case mothershad an overall lower education level and were more likely

to have a history of alcohol drinking compared to controlmothers (Table 1 ). There was no significant difference incases and controls with respect to maternal smokinghabits. Fathers of cases tended to be less educated,

younger at the time of birth of the index child, and more



Table 2 Risk of infant leukemia associated with paternal preconception

.‘ Frequencies were obtained for all cases and controls pooled, ignoring

matching status. Subjects with missing values were excluded. ‘� ORs were

derived from conditional logistic regression model and adjusted for paternalage, education, and drinking.

Cancer Epidemiology, Biomarkers & Prevention 647

likely to be smokers and drinkers than were controlfathers, although most differences were not statistically

significant (Table 1 ). Cases and controls did not differ

appreciably in annual family income. Of the above fac-

tors, parental age, education, and drinking habits were

found to confound the association between radiation andleukemia risk. They were, therefore, adjusted for in the

remaining analyses.

Paternal Preconception X-Ray Exposure. Risk of leukemiawas significantly increased among those children whose

fathers reported ever having pre-conception X-ray exposure

of the chest (OR, 1 .44), limb (OR, 1 .46), upper GI tract (OR,

1 .87), or lower GI tract (OR, 2.24). The elevated risk tendedto be higher for exposures that occurred in the month or the

year before conception than for those occurring earlier,although most point estimates were not statistically signifi-

cant due to the small number of exposed subjects (Table 2).X-ray exposures ofthe head and neck (mainly dental X-rays)

were not related to risk.

The relationship of paternal preconception X-ray ex-

posure to leukemia risk was further evaluated according to

the number of X-ray exposures at specific sites and by majorhistopathological types of leukemia (Table 3). The risk of

ALL increased with the number of X-rays the father received

to the upper GI (trend test, P = 0.04), lower GI and lower

abdomen (P < 0.01 ), and chest (P = 0.08). Risk of ALL wasalso substantially elevated (OR, 2.48; 95% CI, 0.85-7.29),

although not statistically significant, among children whose

fathers had five or more X-rays of the back and spine. Incontrast, risk of ALL was not associated with paternal X-ray

exposure of the head and neck or limbs. The risk of AML

was less consistently associated with paternal X-ray expo-

sure, with an indication of a positive association for X-rays

of the upper Cl (trend test, P = 0.07) but not for otheranatomic locations (Table 3).

Analyses restricted to subjects with direct father inter-view revealed that the ORs of ALL related to lower GI and

abdomen X-ray were increasingly higher (OR, 3.9; 95% CI,1 .77-8.58 and OR, 5.64; 95% CI, 1 .52-20.95 for one andtwo or more exposures, respectively). The ORs of ALL re-lated to upper GI X-ray (OR, 1 .33; 95% Cl, 0.65-2.7 and

OR, 2.15; 95% CI, 0.67-6.96 for one and two or moreexposures) and chest X-ray (OR, 1 .38; 95% Cl, 0.54-3.5 7for 1 0 or more exposures), however, were attenuated.No sign ificant association between parental preconceptionX-ray exposure and risk of AML was observed (data not

shown).Given the face that X-ray exposures at various ana-

tomic sites were correlated, additional analyses were per-formed to adjust for lower GI and abdomen X-ray exposure.

The overall pattern presented above remained, althoughboth the point estimates and confidence intervals of ALLrisk associated with highest X-ray exposures in chest (OR,1 .95; 95% CI, 0.86-4.44), upper CI (OR, 1 .51 ; 95% Cl,

0.46-4.98), back and spine (OR, 2.74; 95% Cl, 0.84-8.98), limb (OR, 1 .49; 95% CI, 0.63-3.50), and head and

neck (OR, 0.62; 95% CI, 0.23-1 .68) changed. The OR forALL related to two or more lower GI and abdomen X-ray

exposures ranged from 2.87 (95% CI, 1 .08-7.58) to 4.53(95% Cl, 1 .64-1 2.54) after adjustment for X-ray in other

sites.The association of paternal preconception X-ray expo-

sure did not vary with age at leukemia diagnosis (�6months, 7-1 2 months, and 1 3-1 8 months) or birth weight

X-ray exposure

Cases Controls

Ever had (%�)“ )o/,,)d OR” �S”/�, CI

n=280 n=511

Total X-ray

Never

Ever (prior to pregnancy)

More than a year

Within a year

Within a month

3.2

96.8

50.4

42.1

4.3

3.3

96.7

57.9

36.6

2.2

1.0

1 .08

0.95

1 .32

2.56

0.42-2.81

0.36-2.52

0.49-3.54

0.67-9.75

Head and neck X-ray

Never


More than a year

Within a year

Withinamonth

8.5

91 .5

54.0

35.3

2.2

5.3

94.7

60.6

32.9

1.2

1.0

0.75

0.69

0.89

1.16

0.39-1 .45

0.35-1 .34

0.45-1.78

0.31-4.36

Chest X-ray

Never


Morethanayear

Within a year

Withinamonth

34.4

65.6

57.5

7.0

1.1

41.5

58.5

51.7

6.4

0.4

1.0

1 .44

1.43

1 .43

7.51

1 .04-2.01

1.01-2.01

0.73-2.79

0.69-81.5

Limb X-ray

Never


More than a year

Within a year

Withinamonth

38.0

62.0

55.6

5.7

0.7

45.2

54.8

49.9

4.3

0.6

1.0

1 .46

1 .44

1 .60

2.79

1 .05-2.03

1 .03-2.01

0.75-3.39

0.37-21.15

Back or spine X-ray

Never


More than a year

Withinayear

Within a month

74.1

25.9

23.3

2.2

0.4

72.6

27.4

25.0

2.0

0.4

1 .0

0.99

0.96

1.18

2.05

0.69-1 .42

0.66-1 .41

0.40-3.49

0.12-15.09

Upper gastrointestinal X-ray

Never


More than a year

Within a year

Within a month

79.9

20.1

1 7.9

1.8

0.4

88.2

1 1 .8

1 1 .6

0.2

0

1.0

1 .87

1 .76

6.48

1 .20-2.90

1 .1 3-2.75

0.71-58.92

Lower gastrointestinal or

abdomen X-ray

Never


Morethan a year

Within a year

Within a month

79.0

21 .0

18.0

2.9

0

88.5

1 1 .5

10.9

0.6

0

1 .0

2.24

1.99

5.93

1.44-3.47

1.25-3.16

1.52-23.10

(�3500 and >3500 grams). The small number of subjectfathers who received X-ray exposures, except for dentalX-rays, in the time periods closer to conception of the indexchild (i.e., a year or a month prior to pregnancy) precludedfurther evaluation of potential combined effects of timingand dose of paternal X-ray exposure.

More fathers of cases than fathers of controls re-

ported exposure to radioactive materials in the occupa-tional setting (OR, 1 .7; 95% CI, 1 .07-2.71 ) or had worna radiation badge at work (OR, 2.25; 95% Cl, 1.16-4.37). The association between paternal diagnostic X-ray

and risk of infant leukemia remained after adjustment for




Table 3 Risk of infant leukemia associated with the number of paternal preconception X-ray exposures

Total leukemia (n = 280) ALL (n = 191)No. of X-rays by anatomic site

OR’ (95% CI) OR� (94% Cl) OR�

AML (n = 79)

(95% Cl)

Head and neck

None 1.0 1.0

1-9 0.77 (0.38-1.53) 0.65 (0.27-1.59)

10-19 0.66 (0.32-1.37) 0.50 (0.20-1.25)

20+ 0.97 (0.46-2.05) 0.67 (0.26-1 .73)

Trendtest P=0.66 P=0.55

1.0

1.01

1.29

1 .82

(0.29-3.60)

(0.31-5.31)

(0.45-7.32)

P=0.19

Chest

None 1.0 1.0

1-4 1.35 (0.95-1.93) 0.93 (0.61-1.44)

5-9 1.46 (0.85-2.52) 1.31 (0.67-2.54)

10+ 1.97 (0.99-3.94) 2.21 (1.0-4.90)

TrendIest P=0.02 P=0.08

1.0

2.93

2.35

1.08

(1.42-6.04)

(0.85-6.51)

(0.15-7.83)

P=0.10

Limb

None 1.0 1.0

1-4 1 .41 (0.99-2.02) 1 .65 (1 .08-2.54)

5-9 1 .78 (1 .05-3.00) 1 .74 (0.92-3.27)

10+ 1.21 (0.61-2.41) 1.41 (0.65-3.05)

Trend test P = 0.07 P = 0.08

1.0

0.99

1 .83

0.77

(0.47-2.06)

(0.67-5.04)

(0.14-4.28)

P = 0.57

Back and spine

None 1.0 1.0

1-2 0.83 (0.54-1.26) 0.76 (0.45-1.28)

3-4 1.11 (0.56-2.18) 1.20 (0.53-2.70)

5+ 2.32 (0.91-5.95) 2.48 (0.85-7.29)

Trend test P = 0.35 P = 0.34

1.0

1.26

1.15

1.02

(0.57-2.80)

(0.31-4.31)

(0.09-11.63)

P = 0.67

Upper gastrointestinal

None 1.0 1.0

1 1 .59 (0.97-2.60) 1 .37 (0.73-2.55)

2+ 3.29 (1.34-8.08) 2.71 (0.99-7.44)

Trend test P < 0.01 P = 0.04

1.0

1 .65

5.31

(0.69-3.94)

(0.53-53.40)

P = 0.07

Lower gastrointestinal and abdomen

None 1.0 1.01 2.38 (1.41-4.02) 3.36 (1.69-6.70)

2+ 2.09 (1.01-4.32) 3.78 (1.49-9.64)

Trendtest P<0.01 P<0.01

1.01.02

0.19

(0.38-2.77)

(0.02-1.72)

P=0.25

., ORs were derived from conditional logistic regression model and adjusted for paternal age, education, and drinking.

occupational radiation exposure. The adjusted ORs forALL were 4.02 (95% CI, 1.55-10.44), 2.72 (95% Cl,

0.98-7.55), and 2.1 8 (95% Cl, 0.98-4.48) for the highestexposure categories of lower GI/abdomen, upper GI, and

chest X-ray, respectively.

Maternal Preconception X-Ray. Maternal diagnostic X-rayexamination a month or more before conception of theindex child was not related to the risk of infant leukemia,irrespective of timing or exposure site (Table 4). MaternalX-ray exposure in the month prior to conception of theindex child, however, was related to an increased risk (OR,4.5; 95% CI, 1.05-19.28). Although based on small num-bers of exposed subjects, elevated risks with X-ray exposurewithin a month prior to pregnancy were observed for most

exposure sites (Table 4).Leukemia risk was unrelated to the number of ma-

ternal preconception X-rays, even when examined withinhistopathological types and exposure sites (Table 5).Analyses stratified by age at leukemia diagnosis showed

that two or more maternal lower GI or abdominal X-ray

exposures were related to an elevated risk of infant leu-

kemia (OR, 4.83; 95% Cl, 1.75-13.38) in children aged7-12 months but not in other infants (<7 months or

1 3-1 8 months; data not shown). Such an excess of risk is

likely due to chance.

Mothers of cases were more likely, although not sta-

tistically significantly, than control mothers to report ever

having been exposed occupationally to radioactive materi-als (OR, 1 .82; 95% Cl, 0.93-3.56). No difference, however,

was observed for cases and controls with respect to mothersreporting use of radiation badges on the job (OR, 1 .05; 95%

Cl, 0.34-3.18).The potential confounding effect of paternal X-ray ex-

posure on maternal exposure and the potential interactive

effect were also evaluated. We did not observe any appre-ciable confounding effect of paternal exposure on maternalexposure or vice versa. There was no indication of interac-tion between paternal and maternal X-ray exposure (datanot shown).

Maternal Radiation Exposure during Pregnancy. Similarproportions of leukemia case mothers (19.5%) and controlmothers (1 8%) reported ever having any X-rays during preg-

nancy. A non-significant excess of in utero X-ray exposure

was found in AML children (OR, 1 .58; 95% CI, 0.80-3.12)but not in ALL cases. X-ray exposure of the upper body and

upper abdomen, which was mainly dental care related, was



Table 4 Risk of infant leukemia associated with maternal preconception

X-ray exposure

Cases ControlsEver had (%)� (%)� OR” 95% CI

n=302 n=558

Discussion

This case-control study was designed to investigate pa-

rental exposures prior to and during pregnancy in the

etiology of infant leukemia. The results showed that pa-

ternal exposure to radiation prior to conception, either as

a medical procedure or occupation related, is associatedTotal X-ray

Never 2.0 1.6 1.0 with an increased risk of leukemia in the first 18 monthsEver (prior to pregnancy) 98.0 98.4 0.87 0.29-2.68 of life. The elevated risk were mainly attributable to

More than a year 51 .3 56.1 0.84 0.27-2.61

Within a year 41.1 40.9 1.02 0.32-3.21

Within a month 5.6 1 .4 4.50 1 .05-19.28

the excess risk of ALL, and the positive association wasmore evident for exposures of high frequency, occurring

at sites closer to the gonads or time periods closer toHead and neck X-ray conception.

Never 4.0 3.2 1 .0

Ever (prior to pregnancy) 96.0 96.8 0.82 0.37-1 .81

More than a ‘ear 56.3 58.2 0.77 0.35-1 .72

Within a year 35.3 37.5 0.84 0.37-1 .91

Within a month 4.3 1.1 3.64 1.04-12.71

The association between paternal preconception radi-

ation exposure and risk of infant leukemia is intriguing.

Alternative explanations, however, need to be considered

before any etiological connection can be made. Of major

Chest X-ray

Never 44.4 43.7 1.0

Ever (prior to pregnancy) 55.6 56.3 1 .04 0.77-1 .40

More than a year 49.2 50.8 1.01 0.74-1.38

concern is the accuracy of exposure assessment, since only

interview information was obtained. Although the short

time period between conception and diagnosis of disease in

infants substantially reduced the likelihood of recall bias,Within a year 5.8 5.3 1.06 0.55-2.06 we can not exclude the possibility that parents of casesWithin a month 0.7 0.2 4.42 0.39-50.60 might recall exposures more readily than parents of controls

Limb X-ray and might telescope the time of exposures. The presence ofNever 61.9 60.2 1.0 paternal X-ray association in ALL but not in AML and theEver (prior to pregnancy) 38.1 39.8 0.95 0.71-1 .28 general absence of association for maternal X-ray exposure,

More than a year 35.1 36.6 0.95 0.70-1 .30

Within a year 2.3 2.7 0.87 0.33-2.27

Within a month 0.7 0.5 1 .44 0.23-8.84

however, argues against information bias as the sole expla-

nation for the data presented. In addition, it seems highlyunlikely that fathers ofALL cases would selectively recall or

Back or spine X-ray

Never 81.9 78.8 1.0

Ever (prior to pregnancy) 18.1 21.2 0.87 0.60-1.25

More than a year 1 5.7 1 9.6 0.83 0.56-1 .21

Within a year 1.7 1.3 1.27 0.38-4.25

Within a month 0.7 0.4 1.97 0.27-14.63

over-report their X-ray exposure according to distance of

exposed sites from the gonads.The two major cooperative clinical trials groups

identify approximately 99% of the expected number of

childhood cancers in the 0-4 year age group (1 1). CCG

Upper gastrointestinal X-ray

Never 80.5 79.2 1 .0

Ever (prior to pregnancy) 19.5 20.8 0.99 0.68-1.45

More than a year 18.9 19.5 1.01 0.69-1.49

Within a year 0.7 1.3 0.54 0.11-2.75

institutions identify approximately one-half of the ex-

pected cases. There is no reason to believe that referralpatterns for infant leukemia cases to CCG versus thePediatric Oncology Group member institutions wouldresult in referral bias. Selective participation of subjects,

Within a month 0 0 however, is of some concern. About 8% of the mothers of

Lower gastrointestinal orabdomen X-ray

eligible cases and 25% of the mothers of eligible controlsrefused to participate in this study. Interviews of the

Never 78.0 76.6 1.0 fathers were not carried out for 27% and 31 % of eligibleEver (prior to pregnancy) 22.0 23.4 1 .00 0.70-1 .42 cases and controls, respectively. It is noteworthy that

More than a year 20.3 22.3 0.99 0.69-1.41

Within a year 1.7 1.1 1.27 0.35-4.58

Within a month 0 0

nonparticipating parents were less educated compared toparticipating parents. To control for this potential selec-tion bias, we adjusted for parental education throughoutour analyses. Residual bias, if any, resulting from selec-

tive participation according to educational level, wouldtend to lead to an underestimation of the disease-radia-

tion association, since high education was found to berelated to more X-ray exposure (12).

Another alternative explanation is that the underlying

a Frequencies were obtained for all cases and controls pooled, ignoring

matching status. Subjects with missing values were excluded.b ORs were derived from conditional logistic regression model and adjusted

for maternal age, education, and drinking.

associated with a marginally elevated risk of AML (Table 6). diseases that required X-ray examination or their relatedPelvimetry and X-rays of the lower abdomen during preg- medications might cause mutation of the parental germnancy were rare in the study population (2.5 and 2.4% of cells and, in turn, increase the risk of leukemia in theirmothers of ALL cases and their matched controls, 2.3 and offspring. To address this concern, we reviewed all ques-1 .9% of mothers of AML cases and their matched controls) tionnaires to check whether the excess lower GI and abdo-and produced ORs of 1 .1 2 and 1 .48 for ALL and AML, men X-ray exposure among cases was attributed to anrespectively (Table 6). excess of certain diseases or medications. We did not ob-

Sixty-eight O/,� of cases and 64.8% of controls had in serve any pattern of diseases or medications that couldutero ultrasound exposure. Neither the cumulative expo- account for the excess of X-rays among cases. Therefore, ifsure nor the timing of ultrasound examination was associ- an underlying disease or medication was the cause of infantated with the risk of ALL or AML, individually or combined leukemia, we would have to assume that many paternal(Table 6). diseases or medications were involved. Unfortunately, a





Table 5 Risk of i nfant leukemia associated with the num ber of mate rnal preconception X-ray exposures

No. of X-rays by anatomic siteTotal leukemia (n = 302)

OR”

ALL (n = 203)

(95% CI)

AML (n = 88)

OR” (95% Cl) OR’ (95”/o Cl)

Head and neck

None

1-9

10-19

20+

Trend test

1.0

0.84 (0.38-1.87)

0.73 (0.73-1.67)

0.86 (0.37-2.00)

P = 0.88

1.0

0.66

0.48

0.61

(0.25-1.73)

(0.18-1.32)

(0.22-1 .69)

P = 0.40

1.0

1.27

1.63

1 .34

(0.24-6.79)

(0.29-9.23)

(0.24-7.55)

P = 0.75

Chest

None

1-4

5-9

10+

TrendIest

1.0

1.01 (0.74-1.39)

1.16 (0.62-2.18)

1.28 (0.58-2.83)

P=0.53

1.0

1.00

1.15

0.70

(0.68-1.48)

(0.53-2.47)

(0.19-2.51)

P=0.92

1.0

1.09

1.22

1.77

(0.60-1.97)

(0.39-3.89)

(0.57-5.56)

P=0.35

Limb

None

1-4

5-9

10+

Trend test

1.0

0.93 (0.68-1.27)

1.00 (0.47-2.13)

1.87 (0.71-4.90)

P = 0.64

1.0

0.89

1.01

2.28

(0.61-1.31)

(0.42-2.40)

(0.66-7.85)

P = 0.68

1.0

0.92

0.82

1.11

(0.52-1.62)

(0.15-4.61)

(0.22-5.71)

P = 0.87

Back and spine

None

1-2

3-4

5+

Trend lest

1.0

0.86 (0.58-1.30)

1.19 (0.52-2.73)

0.76 (0.25-2.30)

P = 0.63

1.0

0.78

1.02

0.63

(0.48-1.27)

(0.36-2.88)

(0.18-22.5)

P = 0.34

1.0

0.93

1.76

1.25

(0.41-2.09)

(0.41-7.66)

(0.11-14.48)

P = 0.65

Upper gastrointestinal

None

1

2+

Trend test

1.0

0.97 (0.64-1 .48)

1.05 (0.51-2.17)

P = 0.99

1.0

0.82

1.17

(0.49-1 .38)

(0.47-2.96)

P = 0.82

1.0

1 .38

0.18

(0.63-2.99)

(0.02-1.56)

P = 0.49

Lower gastrointeslinal and abdomen

None

1

2+

Trend lest

1.0

0.93 (0.61-1.43)

1.13 (0.67-1.90)

P = 0.79

1.0

0.88

1.04

(0.53-1.47)

(0.55-1.97)

P = 0.90

1.0

1.05

1.26

(0.44-2.48)

(0.50-3.21)

P = 0.65

a ORs were derived from conditional logistic regression model and adjusted for maternal age, education, and drinking.

detailed medical history of parents prior to conception of

the index child was not obtained, which precluded a thor-ough evaluation of the confounding effect of medical con-

ditions on X-ray exposure. Nevertheless, the association

between paternal occupational radiation exposure prior toconception and risk of infant leukemia would argue againstpaternal disease history and medication use as a soleexplanation of the radiation-leukemia association that weobserved.

Although the underlying mechanism for the associ-ation of infant leukemia with paternal preconception

diagnostic X-ray is unclear, there are some epidemiolog-ical and biological data available which provide an av-

enue for speculation. Exposure of bone marrow stemcells to X-rays or a-particles has been shown to be relatedto about 2 and 40%, respectively, of chromosome ab-normalities in the daughter cells (13). A delayed chro-mosomal instability was found in cultured cells several

generations after X-ray exposure (1 4, 1 5). Experiments on1 5,000 mice demonstrated that X-ray exposure of paren-tal gonads substantially increased the incidence rate ofboth lung cancer and leukemia in offspring (16). Theseresults indicate a potential role for germline transmission

of mutation or genomic instability in the development ofcancers. The dose of radiation used in those experiments,however, was much higher than that used for humanX-ray examinations. In a United States case-control

study, increased risk of childhood leukemia was associ-ated with maternal (OR, 1 .6) and paternal (OR, 1 .3)preconception X-ray exposures (6). A Chinese study

found a 4-fold elevated risk of leukemia in relation to 10or more paternal preconception X-ray examinations (7).A positive association for maternal exposure was alsonoted but disappeared after adjustment for paternal ex-

posure (7). With documented exposure data, British re-searchers reported a 6-fold risk of childhood acute leu-kemia (mainly ALL) in the offspring of fathers who hadreceived cumulative doses of ionizing radiation greater

than 100 mSv at the nuclear reprocessing plant at

Sellafield (1). A recent reevaluation of a sub-group ofcases who were resident in Seascale near Sellafield atdiagnosis disclosed that all four acute lymphocytic eu-kemia cases with a higher paternal occupational radia-tion exposure (>50 mSv) were diagnosed before age 6 (2

cases were age 2, 1 age 4, and 1 age 5 at diagnosis ofleukemia), while only 1 of the 4 ALL cases with very low




Table 6 Risk of infant leukemia associated with in uter o radiation exposure

Type of exposureTotal cases (n = 302)

OR”

ALL (n = 203)

(95% Cl) OR”

AML (n = 88)

(95% Cl)OR” (95% CI)

Any X-ray

No 1.0 1.0 1.0

Yes 1.12 (0.77-1.63) 0.84 (0.52-1.35) 1.58 0.80-3.12)

Upper body and upper abdomen X-ray

No 1.0 1.0 1.0

Yes 1.10 (0.75-1.62) 0.80 (0.49-1.31) 1.60 (0.80-3.21)

Lower abdomen X-ray and pelvimetry

No 1.0 1.0 1.0

Yes 1.26 (0.48-3.29) 1.12 (0.36-3.50) 1.48 (0.23-9.52)

Ultrasound

No 1.0 1.0 1.0

Yes 1.15 (0.83-1.59) 1.18 (0.79-1.74) 0.99 (0.53-1.86)

1 1.16 (0.81-1.66) 1.24 (0.81-1.91) 0.92 (0.45-1.89)

2 1.20 (0.77-1.86) 1.22 (0.70-2.13) 0.99 (0.46-2.14)

3+ 1.02 (0.59-1.77) 0.89 (0.44-1.79) 1.29 (0.50-3.35)

Trend test P = 0.67 P = 0.88 P = 0.67

Timing of ultrasound

None 1.0 1.0 1.0

1st trimester only 1.01 (0.61-1.68) 0.93 (0.51-1.72) 1.12 (0.39-3.18)

2nd trimester only 1 .04 (0.69-1 .59) 1 .20 (0.72-2.00) 0.70 (0.30-1 .62)

3rd trimester only 1.38 (0.84-2.27) 1.30 (0.70-2.41) 1.07 (0.40-2.83)

1st and 2nd trimesters 1.09 (0.53-2.24) 1.05 (0.40-2.76) 1.07 (0.34-3.39)

1st and/or 2nd and 3rd trimesters 1.12 (0.72-1.74) 1.17 (0.67-2.03) 1.08 (0.50-2.33)

., ORs were derived from conditional logistic regression model and adjusted for maternal age, education, and drinking.

(0.1-49 mSv; 1 case) or no (3 cases) paternal radiationexposure were under age 6 at diagnosis (8). This findingsuggested that paternal preconception radiation exposuremight be responsible for the excess of leukemia amongyoung children, while other risk factors, yet to be iden-tified, might account for leukemia in older children. In

support of this, another United Kingdom-based study

found that paternal preconception occupational expo-sure to radiation, even at doses lower than the 5 mSvlevel, was associated with an increased risk of leukemia

among children under the age of 5 (1 7). A recent Chinesestudy of childhood cancer with direct interview of fathers

of subjects found that paternal diagnostic X-ray exposure2 years prior to conception was related to an elevated riskof cancer among children under age 2 (ORs were 1 .69

and 1 .78, respectively, for 1 and more than 1 X-rayexposure). The excess risk was greater for acute leuke-mia, although the estimates of risk for sub-groups ofcancer were not significant and unstable due to the smallsample size. Paternal diagnostic X-ray exposure in othertime periods and maternal preconception diagnostic X-

ray exposure at any period were not related to the risk of

childhood cancer, either among very young children (�2years of age) or older children (ages 3-14) (18). In con-trast, no elevation in risk of childhood leukemia has been

observed among the offspring of survivors of the Japanese

bombing (2), Canadian nuclear workers (4), or employeesin the Dounreay nuclear plants (2). These studies, how-ever, were limited by low statistical power (2-3), andfailure to adjust for postnatal diagnostic X-ray or envi-ronmental radiation exposure (2-4).

The absence of an association between in uteroX-ray exposure and risk of leukemia in infants, in contract

to the 50-60% increased risk among children under age

15 noted by most previous studies (6-7, 19-21), is un-

expected. However, if the latent period for leukemiaattributed to in utero radiation exposure were longer than

1 8 months, as suggested by the experience of Japanese

atomic bomb survivors (approximately 3-5 years) (22),no excess risk of infant leukemia could be attributed to

prenatal radiation exposure. On the other hand, infantleukemia is different from leukemia in older children in

morphology, clinical presentation (23), and cytogenetic

and molecular findings (24, 25). Infant leukemia, there-fore, may have a distinct etiology. The increased cautionin the use of X-ray examinations during pregnancy andthe reduction over the years in the dose used may beanother alternative explanation.

In summary, this study found that paternal precon-ception exposure to radiation was associated with anincreased risk of infant leukemia. Further studies arewarranted to confirm our finding. Direct evidence linkinggenetic markers to preconceptional exposures are crucial

for resolving the current controversy over germline trans-mission of leukemia and should be a priority for future

studies.

AcknowledgmentsWe would like to acknowledge Dr. Joseph Neglia for help in reviewing theX-ray-related medical conditions for this study and Catherine Moen for help

in preparation of the manuscript.




Appendix Participating principal investigators: Childr ens Cancer Group

Institution Investigator Grant No.

Group Operations Office W. Archie Bleyer, M.D. CA 13539

National Childhood Cancer Foundation, Arcadia, California Anita Khayat, Ph.D.

Harland Sather, Ph.D.

Mark Krailo, Ph.D.

Jonathan Buckley, M.B.B.S., Ph.D.

Daniel Stram, Ph.D.

University of Michigan Medical Center, Ann Arbor, Michigan Raymond Hutchinson, M.D., CA 02971

University of California Medical Center, San Francisco, California Katherine Mattay, M.D. CA 17829

University of Wisconsin Hospital, Madison, Wisconsin Paul Gaynon, M.D. CA 05436

Children’s Hospital and Medical Center, Seattle, Washington James Miser, M.D. CA 10382

Rainbow Babies and Children’s Hospital, Cleveland, Ohio Susan Shurin, M.D. CA 20320

Children’s Hospital National Medical Center, Washington, D.C. Gregory Reaman, M.D. CA 03888

Children’s Memorial Hospital, Chicago, Illinois Edward Baum, M.D. CA 07431

Children’s Hospital of Los Angeles, Los Angeles, California Jorge Ortega, M.D. CA 02649

Children’s Hospital of Columbus, Columbus, Ohio Frederick Ruymann, M.D. CA 03750

Columbia Presbyterian College of Physicians and Surgeons, New York, New York Sergio Piomelli, M.D. CA 03526

Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania Joseph Mirro, M.D. CA 36015

Vanderbilt University School of Medicine, Nashville, Tennessee John Lukens, M.D. CA 26270

Doernbecher Memorial Hospital for Children, Portland, Oregon Robert Neerhout, M.D. CA 26044

University of Minnesota Health Sciences Center, Minneapolis, Minnesota Williams Woods, M.D. CA 07306

Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania Anna Meadows, M.D. CA 1 1 796

Memorial Sloan-Kettering Cancer Center, New York, New York Peter Steinherz, M.D. CA 42764

James Whitcomb Riley Hospital for Children, Indianapolis, Indiana Philip Breitfeld, M.D. CA 13809

Hospital for Sick Children, Toronto, Ontario, Canada Mark Greenberg, MBCHB

University of Utah Medical Center, Salt Lake City, Utah Richard O’Brien, M.D. CA 10198

Strong Memorial Hospital, Rochester, New York Harvey Cohen, M.D. CA 1 1 174

University of British Columbia, Vancouver, British Columbia, Canada Christopher Fryer, M.D. CA 29013

Children’s Hospital Medical Center, Cincinnati, Ohio Robert Wells, M.D. CA 26126

Harbor/UCLA & Miller Children’s Medical Center, Torrance and Long Beach, California Jerry Finkelstein, M.D. CA 14560

University of California Medical Center, Los Angeles, California Stephen Feig, M.D. CA 27678

University of Iowa Hospitals and Clinic, Iowa City, Iowa Raymond Tannous, M.D. CA 29314

Children’s Hospital of Denver, Denver, Colorado Lorrie Odom, M.D. CA 28851

Mayo Clinic, Rochester, Minnesota Gerald Gilchrist, M.D. CA 28882

Izaak Walton Killam Hospital for Children, Halifax, Nova Scotia, Canada Allan Pyesmany, M.D.

University of North Carolina, Chapel Hill, North Carolina Herbert Cooper, M.D.

University of Medicine & Dentistry of New jersey, Camden, New Jersey Milton Donaldson, M.D.

Children’s Mercy Hospital, Kansas City, Missouri Arnold Freeman, M.D.

University of Nebraska Medical Center, Omaha, Nebraska Peter Coccia, M.D.

Wyler Children’s Hospital, Chicago, Illinois F. Leonard Johnson, M.D.

New York University Medical Center, New York, New York Aaron Rausen, M.D.

Children’s Hospital of Orange County, Orange, California Mitchell Cairo, M.D.

References1 . Gardner, M. J., Snee, M. P., HaIl, A. J., Powell, C. A., Downes, S., andTerrell, J. D. Results of the case-control study of leukemia and lymphomaamong young people near Sellafield nuclear plant in West Cumbria. Br. Med.

I., 300:423-429, 1990.

2. Yoshimoto, Y. Cancer risk among children of atomic bomb survivors: a

review of RERF epidemiologic studies. JAMA, 264: 596-600, 1990.

3. Urquhart, J. D., Black, R. J., Muirhead, M. J., et al. Case-control study ofleukaemia and non-Hodgkin’s lymphoma in children in Caithness near the

Dounreay nuclear installation. Br. Med. J., 302: 687-692, 1991.

4. McLaughlin, J. R., Anderson, T. W., Clarke, E. A., and King, W. Occupa-tional Exposure of Fathers to Ionizing Radiation and the Risk of Leukaemia in

Offspring: A Case-control Study. Ottawa, Canada: Atomic Energy ControlBoard, 1992.

5. Park, L., Craft, A. W., Smith, J., et al. Geographical distribution of pre-conceptional radiation doses to fathers employed at the Sellafied nuclear

installation, West Cumbria. Br. Med. J., 307:966-971, 1993.

6. Graham, S., Levin, M. L., Lilienfeld, A. M., et a!. Preconception, intra-uterine, and postnatal irradiation as related to leukemia. NatI. Cancer Inst.

Monogr., 19:347-371, 1966.

7. Shu, X. 0., Gao, Y. T., Brinton, L. A., et a!. A population-based case-

control study of childhood leukemia in Shanghai. Cancer (Phila.), 62:635-644, 1988.

8. Kinlen, L. J. Can paternal preconceptional radiation account for the in-crease of leukaemia and non-Hodgkin’s lymphoma in Seascale? Br. Med. J.,306: 1718-1721, 1993.

9. Robison, L. L., and Daigle, A. E. Control selection using random digitdialing for cases of childhood cancer. Am. J. Epidemiol., 12: 164-166,

1984.

10. Breslow, N. E., and Day, N. E. Statistical Methods in Cancer Research,IARC Scientific Pub. No. 32, Vol. I. Lyon, France: International Agency forResearch on Cancer, 1980.

1 1 . Ross, J. A., Severson, R. K., Robison, L. L., et a!. Pediatric cancer in theUnited States: A preliminary report of a collaborative study of the Childrens

Cancer Group and the Pediatric Oncology Group. Cancer (Phila.), 71(Suppl.): 3415-3421, 1993.

1 2. United States Department of Health, Education, and Welfare. PopulationExposure to X-rays U.S., 1970. DHEW Publication (FDA) 73-8047. Wash-

ington DC: United States Government Printing Office, 1973.

1 3. Kadhim, M. A., Macdonald, D. A., Goodhead, D. T., Lorimore, S. A.,Marsden, S. J., and Wright, E. G. Transmission of chromosomal instability

after plutonium a-particle irradiation. Nature (Lond.), 355: 738-740,

1992.

14. Holmberg, K., Fall, S., Johansson, A., and Lambert, B. Clonal chromo-

some aberrations and genomic instability in X-irradiated human T-Iympho-

cyte cultures. Mutat. Res., 286: 321-330, 1993.

15. Marder, B. A., and Morgan, W. F. Delayed chromosomal instability

induced by DNA damage. Mol. Cell. Bio., 13: 6667-6677, 1993.

16. Nomura, T. Role of radiation-induced mutations in multi-generationcarcinogenesis. In: N. P. Napalkov, J. M. Rice, L. Tomatis, and H. Yamasaki(eds.), Perinatal and Multigeneration Carcinogenesis. IARC Scientific Pub.No. 375. Lyon, France: International Agency for Research on Cancer, 1989.




1 7. Roman, E., Watson, A., Beral, V., et a!. Case-control study of Ieukaemia

and non-Hodgkin’s lymphoma among children aged 0-4 years living inWest Berkshire and North Hampshire health districts. Br. Med. J., 306:615-621, 1993.

1 8. Shu, X. 0., Jin, F., Linet, M. S., et al. Diagnostic X-ray and ultrasoundexposure and risk of childhood cancer. Br. J. Cancer, 70: 531-536, 1994.

19. Stewart, A., Webb, K., and Giles, D. Malignant disease in childhood and

diagnostic irradiation in utero. Lancet, 2: 447, 1956.

20. Gilman, E. A., Kneale, G. W., Knox, E. G., and Stewart, A. M. Pregnancyx-rays and childhood cancers: effects of exposure age and radiation dose.J. Radiol. Prot., 8:3-8, 1988.

21 . Mole, R. H. Childhood cancer after prenatal exposure to diagnosticx-ray examination in Britain. Br. J. Cancer, 62: 1 52-1 60, 1990.

22. Finch, S. C. Leukemia and lymphoma in atomic bomb survivors. In:

J. D. Boice, Jr. and J. D. Fraumeni, Jr. (eds.), Radiation Carcinogenesis: Epi-demiology and Biological Significance. New York: Raven Press, 1984.

23. Katz, F., Ball, S., and Gibbons, B. Acute lymphoblastic leukaemia ininfancy: clinical and biological features. Leuk. Lymphoma, 2: 259-269,1990.

24. Chen, C-H., Sorensen, P. H. B., Domer, P. H., et a!. Molecular rear-rangements on chromosome 1 1q23 predominate in infant acute lympho-

blastic leukemia and are associated with specific biologic variables and pooroutcome. Blood, 81: 2386-2393, 1993.

25. Djabali, M., Selleri, L., Parry, P., Bower, M., Young, B. D., and Evans,

G. A. A trithorax-like gene is interrupted by chromosome 1 1q23 transloca-tions in acute leukemias. Nature Genet., 2: 13-18, 1992.



1994;3:645-653. Cancer Epidemiol Biomarkers Prev X O Shu, G H Reaman, B Lampkin, et al. infant leukemia. Investigators of the Childrens Cancer Group.Association of paternal diagnostic X-ray exposure with risk of

Updated version

http://cebp.aacrjournals.org/content/3/8/645

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cebp.aacrjournals.org/content/3/8/645To request permission to re-use all or part of this article, use this link



http://cebp.aacrjournals.org/cgi/alerts

mailto:[email protected]



Association of Paternal Diagnostic X-Ray Exposure with ... · leukemia, higher risks being linked to exposures closer to conception. X-ray related leukemia risk varied with exposure

Documents