Parental medication use and risk of childhood acute lymphoblastic leukemia

Authors


Abstract

BACKGROUND

Few studies have examined the risk of childhood acute lymphoblastic leukemia (ALL) associated with parental medication use. As part of a large case–control study conducted by the Children's Cancer Group, we evaluated the association between maternal and paternal medication use and the risk of ALL in offspring.

METHODS

Information on selected medication use in the year before and during the index pregnancy was obtained by telephone interview. Participants included 1842 children of 14 years or younger with newly diagnosed and immunophenotypically defined ALL and 1986 individually matched controls. Data were analyzed using logistic regression models and stratified by immunophenotypes of ALL and age at diagnosis of cases.

RESULTS

After adjusting for potential confounders and other medication use, we found that maternal use of vitamins (odds ratio [OR] = 0.7, 99% confidence interval [CI]: 0.5–1.0) and iron supplements (OR = 0.8, 99% CI: 0.7–1.0) only during the index pregnancy was associated with a decreased risk of ALL. Parental use of amphetamines or diet pills and mind-altering drugs before and during the index pregnancy was related to an increased risk of childhood ALL, particularly among children where both parents reported using these drugs (OR = 2.8, 99% CI: 0.5–15.6 for amphetamines or diet pills, OR = 1.8, 99% CI: 1.1–3.0 for mind-altering drugs). Stratified analyses showed that maternal use of antihistamines or allergic remedies and parental use of mind-altering drugs were strongly associated with infant ALL, whereas patterns of association between childhood ALL and parental medication use did not influence markedly the immunophenotypic subgroup of ALL.

CONCLUSIONS

The findings of this study suggest that certain parental medication use immediately before and during the index pregnancy may influence risk of ALL in offspring. Cancer 2002;95:1786–94. © 2002 American Cancer Society.

DOI 10.1002/cncr.10859

The effect on the health of the offspring of parental use of medications or drugs immediately before or during pregnancy is a continuing public health concern. Published data1 indicate that some medications are used more frequently during pregnancy than was expected. Maternal medication use is associated with adverse pregnancy outcomes, including congenital malformations,2, 3 low birth weight,4, 5 and both physical and developmental delays in offspring.6–8

Acute lymphoblastic leukemia (ALL), the most common childhood malignancy worldwide, peaks at ages 2–5,9, 10 indicating a possible causative effect of in utero or early life exposure. The in utero origin of ALL is supported by recent studies,11, 12 further emphasizing the importance of early exposure.

There are limited data that address the potential association between maternal medication use and childhood leukemia. One study reported that maternal use of marijuana and antinausea medication was associated with an increased risk of childhood acute myeloid leukemia (AML).13 Several studies comprising adults reported the possible association between self-prescribed medication and non-Hodgkin lymphoma,14, 15 leukemia,15–17 and renal cell carcinoma.18 However, the findings have not been consistent and no study has examined specifically the associations between parental medication use and the risk of childhood ALL.

The Children's Cancer Group (CCG) completed a large-scale case–control study to evaluate the potential risk factors for childhood ALL. As part of this study, we report the results of the analyses addressing parental medication use and the risk of ALL.

MATERIALS AND METHODS

The CCG was a National Cancer Institute-funded cooperative clinical trials group (now encompassed by the Children's Oncology Group) with 118 members and affiliated institutions in the United States, Canada, and Australia. Cases in this study were identified through CCG institutions in the United States and Canada.19 Eligible cases were newly diagnosed with ALL between January 1, 1989, and June 15, 1993, were younger than the age of 15 years at diagnosis, and lived in a home with a telephone. Controls were selected randomly using a random digit dialing methodology.20 They were matched individually to cases on age (within 25% of age at diagnosis of the case, with a maximum difference of ±2 years), race (white, black, or other), and telephone area code and exchange. In a few situations where an exact match could not be achieved after dialing 300 random numbers, the age and race match criteria were relaxed. The control-to-case ratio was generally 1:1, except for the rare T-cell ALL subgroup, in which more than one control was matched to each case. An additional eligibility criterion included the availability of an English-speaking biologic mother for interview.

A standard panel of monoclonal antibodies was used with all bone marrow specimens to determine B-cell or T-cell lineage. A subset of B-lineage leukemias was classified further by the determination of cytoplasmic immunoglobulin. Cases were assigned to one of the following mutually exclusive groups: T-cell ALL, early pre-B ALL (B-lineage markers positive, cytoplasmic immunoglobulin negative), pre-B ALL (B-lineage markers positive, cytoplasmic immunoglobulin positive), B-lineage ALL-not otherwise specified (B-lineage markers positive, but cytoplasmic immunoglobulin not performed), or unclassifiable.

During the study period, 2081 eligible cases and 2597 eligible controls were identified. Data were collected by independent telephone interviews with mothers and, whenever available, fathers of cases and controls, using structured questionnaires. The mother's interview was completed for 1914 cases (92.0%) and 1987 controls (76.5%). Of the 2081 eligible cases, 58 (2.8%) had died. Of the 1914 cases with mother's interview, 27 (1.4%) had died before the time of interview. The 167 nonparticipating cases included 70 (3.4%) parental refusals, 41 (2.0%) physician refusals, 18 (0.9%) cases who were lost to follow-up, and 38 (1.8%) cases who did not participate for miscellaneous reasons. The 610 nonparticipating controls included 457 (17.6%) parental refusals, 17 (0.7%) controls who were lost to follow-up, and 136 (5.2%) controls who did not participate for miscellaneous reasons. The median (range) of the interval between the date of interview and the date of case diagnosis was 228 (0–1345) days. After excluding nonparticipants and unmatched subjects, 1842 case–control sets (1842 cases, 1986 controls, 1704 sets with 1:1 matching, 132 sets with 1:2 matching, and 6 sets with 1:3 matching) remained for statistical analyses. The father's interview was completed for 1800 cases (86.5%) and 1814 controls (69.8%); mothers served as surrogates for 268 (16.6%) case fathers and 557 (32.3%) control fathers, resulting in 1618 case–control sets (1618 cases, 1722 controls).

Self-reported information included demographics, parental history of disease, medication use, occupation, personal habits, household exposure history, maternal pregnancy and birth history, as well as exposure to environmental hazards of the index child. This analysis included information on maternal and paternal use of a variety of medications during the 1 year before the index pregnancy and maternal use during the pregnancy and nursing of the index child. For each type of medication, mothers and fathers were asked to provide the name, total times used, and duration of use of the medications if used for 5 days or more during each of these time periods.

Conditional logistic regression models were used to obtain odds ratios (OR) and corresponding 99% confidence intervals (CI). Crude ORs were first estimated and then ORs were adjusted for potential confounders in multivariate models. Potential confounders were chosen from among demographic characteristics and life-style factors because of their reported associations with ALL risk21 and medication use.22 Because use of various medications might be mutually related, selected medications, together with other confounders, were simultaneously included in the logistic model. Considering multiple tests in this analysis, we chose a more strict standard, P < 0.01, to judge statistic significance.

RESULTS

Table 1 describes the distribution of ALL immunophenotypes and other characteristics of cases and controls. There were more boys among cases (55.3%) and cases were predominantly between the ages of 2 and 5 years (55.4%). Controls were similar to cases on gender and age distribution, but were more likely to come from a family with a higher income and parental education level and were also more likely to be white. Compared with control parents, case mothers were less likely to have reported consuming alcohol before or during the index pregnancy and case fathers were more likely to have smoked (P = 0.03) before or during the index pregnancy. These demographic characteristics and parental smoking and drinking status before or during pregnancy were potential confounders and adjusted for in all subsequent analyses.

Table 1. Characteristics of Cases and Controls
CharacteristicsCases (%)Controls (%)P value/OR (99% CI)
  • CI: confidence interval; OR: odds ratio.

  • a

    Significant at 0.01 level.

Among matched subjects with maternal interview completed (1842 cases, 1986 controls)
 Immunophenotype   
  T cell183 (9.9)  
  Early pre-B cell893 (48.5)  
  Pre-B cell233 (12.6)  
  B not specified231 (12.5)  
  Unknown302 (16.4)  
 Gender   
  Male1018 (55.3)1076 (54.2) 
  Female824 (44.7)910 (45.8)P = 0.50
 Age   
  < 12 mos64 (3.5)81 (4.1) 
  12–23 mos138 (7.5)189 (9.5) 
  2–5 yrs1020 (55.4)1038 (52.3) 
  6–10 yrs408 (22.1)466 (23.5) 
  11–14 yrs212 (11.5)212 (10.7)P = 0.07
 Household income   
  < $20,000607 (33.0)546 (27.5)1.0 (reference)
  $20,000–$39,999767 (41.6)844 (42.5)0.8 (0.6–1.0)a
  ≥ $40,000468 (25.4)596 (30.0)0.6 (0.5–0.8)a
 Maternal race   
  White1544 (83.8)1765 (88.9)1.0 (reference)
  Nonwhite298 (16.2)221 (11.1)2.6 (1.7–4.0)a
 Maternal education   
  High School or lower797 (43.3)762 (38.4)1.0 (reference)
  Some post-high school592 (32.1)701 (35.3)0.8 (0.6–1.0)a
  College or higher453 (24.6)523 (26.3)0.8 (0.6–1.0)a
 Maternal smoking before or during pregnancy
  No1278 (69.8)1418 (71.6)1.0 (reference)
  Yes554 (30.2)562 (28.4)1.1 (0.9–1.3)
 Maternal drinking before or during pregnancy
  No819 (44.5)790 (39.8)1.0 (reference)
  Yes1022 (55.5)1196 (60.2)0.8 (0.7–1.0)a
Among matched subjects with paternal interview completed (1618 cases, 1722 controls)
 Paternal race   
  White1376 (85.0)1542 (89.5)1.0 (reference)
  Nonwhite242 (15.0)180 (10.5)2.2 (1.5–3.4)a
 Paternal education   
  High school or lower676 (41.8)638 (37.0)1.0 (reference)
  Some post high school480 (29.7)510 (29.6)0.9 (0.7–1.1)
  College or higher462 (28.6)574 (33.3)0.7 (0.6–0.9)a
 Paternal smoking before or during maternal pregnancy
  No984 (61.3)1103 (64.4)1.0 (reference)
  Yes621 (38.7)610 (35.6)1.2 (1.0–1.4)
 Paternal drinking before or during maternal pregnancy
  No459 (29.1)471 (28.2)1.0 (reference)
  Yes1120 (70.9)1200 (71.8)1.0 (0.8–1.2)

Table 2 displays the type and frequencies of medications reported ever used by parents during the periods of pre-pregnancy, pregnancy, and nursing of index child and their associations with childhood ALL. After adjusting for the potential confounders listed in Table 1 and other medication use, maternal use of vitamins (the most commonly used medication) was associated with a decreased risk of childhood ALL. Maternal use of antihistamines or allergy remedies and mind-altering drugs was related to an increased ALL risk. Paternal use of amphetamines or diet pills (a slang for amphetamines) and mind-altering drugs were also associated with an increased risk of ALL. Of those reporting use of mind-altering drugs, marijuana was the dominant drug reported by paternal (406 of 451 = 90.0%) and maternal users (189 of 227 = 83.3%). Analysis restricted to marijuana use provided similar results. We found a similar pattern (data not shown) after restricting the analyses to those with direct father interview.

Table 2. Associations between Parental Medication Use and ALL Risk: Adjusted ORs and 99% CI
MedicationsMothersaFathersb
Cases (n = 1842)Controls (n = 1986)OR99% CICases (n = 1618)Controls (n = 1722)OR99% CI
  • ALL: acute lymphoblastic leukemia; OR: odds ratio; CI: confidence interval.

  • a

    Ever used medications during the 1 yr before pregnancy through pregnancy and nursing.

  • b

    Ever used medications during the 1 yr before pregnancy.

  • c

    Significant at 0.01 level.

Oral contraceptives5615691.10.9–1.4/   
Bendectin or other morning sickness tablets1261081.30.9–1.9/   
Sedatives, tranquilizers, or sleeping pills37271.40.7–2.836291.00.5–2.1
Amphetamines or diet pills97751.30.8–2.047202.21.0–4.6c
Vitamins162218270.70.5–1.0c4043951.10.9–1.4
Iron supplements5926970.90.7–1.01171.80.5–6.6
Diuretics or Lasix32261.20.6–2.59120.70.2–2.3
Blood pressure tablets23131.80.7–4.515240.70.3–1.7
Dilantin or other antiepileptic (seizure) drugs670.80.2–3.6717.50.5–122.0
Antibiotics4554761.00.8–1.32182171.10.8–1.5
Prescription cold or cough remedies79860.90.6–1.431341.10.6–2.2
Antihistamines or allergy remedies2231931.31.0–1.8c2032111.00.8–1.4
Mind-altering drugs130971.51.0–2.1c2561951.31.0–1.8c
Hormones84891.00.7–1.61553.30.8–13.2
Laxatives or antidiarrhea agents70531.50.9–2.58150.50.2–1.7
Immunosuppressants or steroids35490.70.4–1.432221.40.7–3.0
Antiinflammatory drugs61521.20.7–2.154640.90.5–1.4
Pain relievers57680.80.5–1.450590.80.5–1.4
Medications not specified4414271.10.9–1.41741850.90.6–1.2

Because maternal use of some medications such as bendectin, vitamins, and iron supplements might be pregnancy related, we further subdivided the maternal data by the specific time periods during which these medications were used. Three mutually exclusive categories of medication use were defined, i.e., 1) used only during 1 year before the index pregnancy, 2) both before and during the index pregnancy, and 3) only during the index pregnancy. As shown in Table 3, the inverse association between ALL risk and maternal use of vitamins was restricted to children whose mothers ever used vitamins during the index pregnancy (adjusted OR = 0.7, 99% CI: 0.5–1.0 for those whose mothers used vitamins both before and during the index pregnancy and for those whose mothers used vitamins only during the index pregnancy). Maternal use of iron supplements only during the index pregnancy was also inversely associated with ALL risk (adjusted OR = 0.8, 99% CI: 0.7–1.0). The risk associated with maternal use of antihistamines or allergy remedies and mind-altering drugs remained increased across all three categories. However, most of the associations were not statistically significant, perhaps because of the decreased sample size due to data stratification.

Table 3. Associations between Maternal Medication Use and ALL risk, by Pregnancy-Related Time Periods: Adjusted ORs and 99% CI
MedicationsOnly before pregnancyBoth before and during pregnancyOnly during pregnancy
Case/controlOR (95% CI)Case/controlOR (95% CI)Case/controlOR (95% CI)
  • ALL: acute lymphoblastic leukemia; OR: odds ratio; CI: confidence interval.

  • a

    Significant at 0.01 level.

Oral contraceptives487/5111.1 (0.9–1.3)69/541.5 (0.9–2.5)2/11.6 (0.1–37.4)
Bendectin or other morning sickness tablets0/2n/a0/1n/a126/1051.4 (0.9–2.0)
Sedatives, tranquilizers, or sleeping pills20/111.7 (0.6–4.8)7/81.3 (0.3–5.2)9/81.2 (0.3–4.4)
Amphetamines or diet pills82/651.3 (0.8–2.0)12/91.4 (0.4–4.8)3/11.9 (0.1–39.3)
Vitamins21/151.2 (0.4–3.0)310/3440.7 (0.5–1.0)a1287/14650.7 (0.5–1.0)a
Iron supplements21/141.4 (0.6–3.6)56/401.5 (0.8–2.6)490/6160.8 (0.7–1.0)a
Diuretics or Lasix13/81.6 (0.5–5.3)2/40.7 (0.1–6.4)15/121.3 (0.4–3.6)
Blood pressure tablets0/1n/a5/23.2 (0.4–28.9)17/91.9 (0.6–5.5)
Dilantin or other antiepileptic (seizure) drugs1/30.4 (0.0–7.8)3/40.7 (0.1–5.7)2/0n/a
Antibiotics177/1860.9 (0.7–1.3)27/310.9 (0.4–1.8)191/2091.0 (0.8–1.3)
Prescription cold or cough remedies26/221.0 (0.5–2.2)3/80.4 (0.1–2.0)47/481.0 (0.6–1.7)
Antihistamines or allergy remedies98/981.1 (0.8–1.7)38/331.3 (0.7–2.6)69/561.4 (0.8–2.2)
Mind-altering drugs70/491.6 (0.9–2.6)50/431.4 (0.8–2.4)10/51.7 (0.4–7.5)
Hormones43/421.2 (0.6–2.1)22/320.7 (0.3–2.4)18/151.5 (0.6–3.7)
Laxatives or antidiarrhea agents11/52.1 (0.5–8.5)6/61.1 (0.2–5.3)42/331.5 (0.8–2.8)
Immunosuppressants or steroids12/140.8 (0.3–2.4)11/101.0 (0.3–3.1)7/210.4 (0.1–1.4)
Antiinflammatory drugs36/371.0 (0.5–2.0)8/33.8 (0.4–32.7)11/71.7 (0.5–6.4)
Pain relievers15/170.8 (0.3–2.1)8/160.5 (0.2–1.7)31/301.1 (0.5–2.1)
Medications not specified105/1101.0 (0.7–1.5)88/781.2 (0.8–1.8)208/2041.2 (0.9–1.5)

Because both maternal and paternal use of amphetamines or diet pills and mind-altering drugs were associated with an increased risk of childhood ALL, we conducted combined analyses of maternal and paternal use of amphetamines or diet pills and mind-altering drugs (Table 4). Compared with children for whom either the mother or father was exposed, children for whom both parents used amphetamines or diet pills (adjusted OR = 2.8, 99% CI: 0.5–15.6) or mind-altering drugs (OR = 1.8, 99% CI: 1.1–3.0) had the highest risk of ALL.

Table 4. Associations between Parental Use of Amphetamines/Diet Pills and Mind-Altering Drugs before and during the Index Pregnancy and ALL Risk: Adjusted ORs and 99% CI
Parental useCasesControlsOR99% CI
  • ALL: acute lymphoblastic leukemia; OR: odds ratio; CI: confidence interval.

  • a

    Significant at 0.01 level.

Mothers    
 Amphetamines or diet pills97751.30.9–1.8
 No. of times used (by median)    
  1–6055381.50.8–2.6
  > 6041371.10.6–2.0
 Mind-altering drugs130971.51.1–2.0a
 No. of times used (by median)    
  1–5574481.71.0–2.8a
  > 5553481.20.7–2.1
Fathers    
 Amphetamines or diet pills47202.21.2–3.9a
 No. of times used (by median)    
  1–801882.10.7–6.6
  > 802382.70.9–8.0
 Mind-altering drugs2561951.31.1–1.7a
 No. of times used (by median)    
  1–66123901.41.0–2.1
  > 66123901.40.9–2.0
Mothers and fathers combined    
 Amphetamines or diet pills    
  Neither149516411.0 (reference) 
  Mothers only36171.20.8–1.9
  Fathers only76612.00.9–4.5
  Both1132.80.5–15.6
 Mind-altering drugs    
  Neither132815021.0 (reference) 
  Mothers only1751431.50.7–3.0
  Fathers only34251.30.9–1.8
  Both81521.81.1–3.0a

The statistically significant associations shown in Tables 2–4 were further evaluated according to the age at case diagnosis and immunophenotype of cases (Table 5). ORs for maternal use of antihistamines or allergic remedies (adjusted OR = 4.3, 99% CI: 0.6–32.1), maternal use of mind-altering drugs (adjusted OR = 6.4, 99% CI: 0.3–127.0), and paternal use of mind-altering drugs (adjusted OR = 3.0, 99% CI: 0.7–12.3) were stronger for infant leukemia cases (diagnosed before the age of 1 year), although none were statistically significant. Cytogenetic data were available for 589 of 1842 ALL cases. Of the 589 cases, the (4;11) translocation was found in 15, the (9;22) translocation in 8, and the (8;14) translocation in 1 case. All 15 (4;11) translocation cases were infants. Of these 15 cases, the mothers of 4 cases reported taking antihistamines or allergy remedies, the fathers of 2 cases took mind-altering drugs, and none of the parents took amphetamines or diet pills. No substantial differences in ALL risk across immunophenotypes were found, except perhaps for differences in risk associated with maternal use of mind-altering drugs.

Table 5. Childhood ALL Risk and Selected Parental Medications, by Age at Diagnosis and Immunophenotypes: Adjusted ORs and 99% CI
Parental medications< 1 yr1–5 yrs6–14 yrsEarly pre-BPre-BT cell
  • ALL: acute lymphoblastic leukemia; OR: odds ratio; CI: confidence interval.

  • a

    Including those used only during 1 yr before the index pregnancy, those used both before and during the index pregnancy, and those used only during the index pregnancy.

  • b

    Including those used both before and during the index pregnancy and those used only during the index pregnancy, excluding those used only during 1 yr before the index pregnancy.

  • c

    Including those used only during the index pregnancy, excluding those used during 1 year before the index pregnancy and those used both before and during the index pregnancy.

  • d

    Significant at 0.01 level.

Mothers      
 Amphetamines or diet pillsa/1.5 (0.9–2.7)1.0 (0.5–2.0)1.2 (0.6–2.2)2.2 (0.6–7.5)1.3 (0.4–4.1)
 Vitaminsb0.9 (0.1–7.1)0.8 (0.5–1.2)0.5 (0.3–0.9)d0.8 (0.5–1.3)0.6 (0.3–1.4)0.3 (0.1–0.8)d
 Iron supplementsc1.1 (0.3–3.9)0.7 (0.6–1.0)0.9 (0.7–1.3)0.8 (0.6–1.1)0.8 (0.5–1.3)0.7 (0.3–1.4)
 Antihistamines or allergy remediesa4.3 (0.6–32.1)1.4 (1.0–2.0)1.1 (0.6–1.8)1.3 (0.9–1.9)2.1 (0.9–4.8)1.5 (0.6–4.1)
 Mind-altering drugsa6.4 (0.3–127.0)1.4 (0.8–2.2)1.4 (0.7–2.6)1.8 (1.0–3.1)d1.3 (0.5–3.6)0.4 (0.1–1.6)
Fathers      
 Amphetamines or diet pills/2.1 (0.8–5.8)2.4 (0.8–7.1)1.9 (0.7–5.4)4.5 (0.6–34.6)2.1 (0.2–21.8)
 Mind-altering drugs3.0 (0.7–12.3)1.2 (0.9–1.7)1.4 (0.9–2.4)1.3 (0.9–2.0)1.5 (0.7–3.4)1.2 (0.4–3.1)

DISCUSSION

In this large case–control study assessing a wide spectrum of self-reported medication use, maternal use of vitamins and iron supplements during pregnancy was associated with a reduced risk of childhood ALL. Vitamin and iron supplementation during pregnancy is a common practice. Although an antileukemia effect has not been reported previously, the benefits of prenatal vitamin and mineral supplementation in improving pregnancy outcomes and reducing the risk of congenital malformations have been well documented.23, 24 Many studies have shown that vitamins, particularly carotenoids, vitamin C, and vitamin E, reduce the risk of lung, breast, esophagus, stomach, colon, uterine cervix, and ovarian carcinomas through their role as antioxidants.25–31 In a large international study of childhood brain tumors including 1051 cases and 1919 controls, Preston-Martin et al.32 reported that maternal vitamin use during pregnancy for two trimesters might decrease the risk (OR = 0.7, 95% CI: 0.5–0.9), with a trend toward lower risk with longer duration of use. We can only speculate that prenatal vitamin supplements exert their possible preventive effect on childhood ALL through possible antioxidative mechanisms, enhancement of immune function,31 or through other mechanisms. For example, most of the vitamin preparations used during pregnancy contain a fair amount of iron content. Unfortunately, our data have no specific information on iron content. Therefore, we could not exclude the possibility that the protective effect of vitamins was confounded by the iron content contained in vitamin preparations.

Maternal use of iron supplements during the index pregnancy was associated with a decreased risk of ALL. This finding is consistent with a recent report, in which Thompson et al.33 reported a protective association between iron or folate supplementation in pregnancy and risk of childhood ALL. However, an increased risk of ALL was observed in children whose mothers ever used iron supplements before the index pregnancy, although this was not statistically significant. The opposite direction of the associations between ALL risk and iron supplements taken before and during pregnancy suggests that the underlying reasons for use of iron supplements during the two time periods may be different. Maternal use of iron supplements before pregnancy may be an indication of anemia. Roman et al.34 reported that mothers of leukemia cases were more likely to be anemic (OR = 3.1, 95% CI: 1.3–11.1). Conversely, iron supplementation during pregnancy, along with vitamin use, is recommended to meet the increased physiologic needs of pregnancy and to prevent anemia. Pregnancy anemia was related to the risk of childhood leukemia in a case–control study conducted by Petridou et al.35 (OR = 2.6, 95% CI: 1.4–4.9). In our study, the mothers of 303 cases and 325 controls had anemia during pregnancy and pregnancy-associated anemia was not associated with ALL risk (OR = 1.0, 99%CI: 0.8–1.3).

A previous CCG study13 showed a 10-fold risk of childhood AML to be associated with maternal marijuana use before and during pregnancy with the index child. Compared with AML cases not exposed to marijuana, exposed cases were significantly younger at diagnosis. Consistent with these findings, our study also found that paternal and maternal use of mind-altering drugs before and during pregnancy increased the risk of childhood ALL in offspring, particularly the risk of infant ALL and particularly when both parents used the drugs. This finding is supported by in vitro and in vivo experimental studies in animals that show that marijuana smoke is mutagenic and fetotoxic.36

Amphetamine use is related to renal cell carcinoma.18, 37 In our study, reported parental, particularly paternal, use of amphetamines or diet pills was associated with an increased risk of childhood ALL. Maternal use of antihistamines or allergy remedies and laxative or antidiarrhea agents, although not statistically significant for the latter, was also associated with increased risk of ALL mainly when used during pregnancy. The risk associated with the use of these medications was stronger among infants. It is unclear whether medication use, underlying diseases, uncontrolled confounding, or chance accounts for the increased risk of ALL. In a case–control study, Cordier et al.,38 reported that antihistamine intake was associated with an increased risk of childhood brain tumors. Laxatives, some of which may inhibit DNA topoisomerase II, have also been hypothesized to be potential leukemogens for specific biologic subgroups of infant leukemias.39 Further studies are needed to establish the safety of these medications during pregnancy.

Strong associations between the risk of infant ALL and maternal use of antihistamines or allergy remedies and parental use of mind-altering drugs, although not statistically significant, were noteworthy. Infant ALL generally shows distinctive biologic features. Cytogenetic and molecular studies indicate that many infant leukemias have chromosome 11q23 translocations.40 All 15 (4;11) translocation cases found in our study were infants. The Children's Oncology Group is currently conducting a large-scale study of infant leukemia including an assessment of whether intrauterine exposure to these and other medications can cause chromosomal damage and lead to the development of infant ALL.

As in many other case–control studies, the interpretation of our findings is hindered by several methodologic limitations. In our analysis, we investigated a relatively large number of medications within numerous exposure periods and conducted multiple tests and comparisons. We employed a stricter standard to judge statistical significance (P < 0.01) to reduce the possibility of chance findings. Information on medication use was self-reported and the data for some control fathers were obtained from mothers who served as surrogates, raising a concern of potential recall or reporting bias.41 The use of amphetamines and mind-altering drugs could be underreported, particularly for the control fathers, which may have resulted in overestimated risks. However, The bidirectional relation between ALL risk and the use of iron supplements before and during pregnancy argues against recall bias as an explanation at least for that finding. Another limitation of this study is that we were unable to distinguish between associations of the risk of ALL with underlying diseases and with relevant medication use. Some medication use could act as a surrogate for parental health status and parental nutrition (e.g., diet pills, vitamin supplements). As Patterson et al.42 suggested, use of vitamins and iron supplements as well as mind-altering drugs may be a behavioral marker for other factors that were related to childhood leukemia risk such as family income, parental race and education, smoking, and drinking habits. Although the potential confounding due to all of these factors was controlled in the analyses, the completeness of adjustment can never be assured. Finally, the lack of a dose-response relationship in maternal use of amphetamines and mind-altering drugs (Table 4) and the wide spectrum of class of chemicals included in these medications raise questions about the nature of some associations found in this study. Our study is one of the few that specifically evaluated relations between parental medication use and the risk of childhood ALL. Our results suggest that maternal use of vitamins and iron supplements during pregnancy, as well as parental use of amphetamines or diet pills and mind-altering drugs, may influence the risk and should be considered for inclusion in any future studies of childhood ALL.

Ancillary