Talc use, variants of the GSTM1, GSTT1, and NAT2 genes, and risk of epithelial ovarian cancer

Abstract

Epidemiologic evidence suggests a possible association between genital use of talcum powder and risk of epithelial ovarian cancer; however, the biologic basis for this association is not clear. We analyzed interactions between talc use and genes in detoxification pathways (GSTM1, GSTT1 and NAT2) to assess whether the talc/ovarian cancer association is modified by variants of genes potentially involved in the response to talc. Our analysis included 1,175 cases and 1,202 controls from a New England-based case-control study and 210 cases and 600 controls from the prospective Nurses' Health Study. We genotyped participants for the GSTM1 and GSTT1 gene deletions and three NAT2 polymorphisms. We used logistic regression to analyze the main effect of talc use, genotype, and gene-talc interactions in each population, and we pooled the estimates using a random effects model. Regular talc use was associated with increased ovarian cancer risk in the combined study population (relative risk=1.36, 95% CI=1.14-1.63; p-trend<0.001). Independent of talc, the genes examined were not clearly associated with risk. However, the talc/ovarian cancer association varied by GSTT1 genotype and combined GSTM1/GSTT1 genotype. In the pooled analysis, the association with talc was stronger among women with the GSTT1-null genotype (p-interaction=0.03), particularly in combination with the GSTM1-present genotype (p-interaction=0.03). There was no clear evidence of an interaction with GSTM1 alone or NAT2. These results suggest that women with certain genetic variants may have a higher risk of ovarian cancer associated with genital talc use. Additional research is needed on these interactions and the underlying biologic mechanisms.

INTRODUCTION

Genital use of talcum powder has been extensively investigated as a potential risk factor for ovarian cancer. A meta-analysis of 16 previous studies reported an approximate 30% increase in risk of total epithelial ovarian cancer with regular genital exposure to talc (1), and several studies have suggested a stronger association with the serous or serous invasive histologic subtype (2-6). Although the epidemiologic evidence supports a modest association between genital talc use and ovarian cancer risk, the association remains controversial due to the lack of a clear dose-response with increasing frequency or duration of talc use, the possibility of confounding or other biases, and the uncertain biologic mechanism.

No prior studies have assessed gene-talc interactions in ovarian cancer risk, possibly because little is known about which genes may be involved in the biologic response to talc. However, variants of the glutathione S-transferase M1 (GSTM1) and N-acetyltransferase 2 (NAT2) genes appear to modify the association between exposure to asbestos, a known carcinogen that is chemically similar to talc, and risk of malignant mesothelioma (7-10). Talc and asbestos are found together in nature, and prior to 1976 talcum powder was commonly contaminated with asbestos (9). Although this contamination may have contributed to the risk of ovarian cancer associated with talc use, there is also evidence that talc itself may contribute to carcinogenesis, independent of any contamination with asbestos in the past. Talc can induce granulomas and other inflammatory responses in vivo (9), and a recent study found that exposing human ovarian stromal and epithelial cells to talc resulted in increased cell proliferation and neoplastic transformation of cells (11). Talc also appears to increase cellular production of reactive oxygen species (11). Interestingly, serous ovarian cancers morphologically resemble peritoneal malignant mesotheliomas (12), suggesting a possible rationale for the stronger association between talc and risk of serous or serous invasive cancers observed in some studies.

Based on similarities between talc and asbestos and the evidence for gene-asbestos interactions in malignant mesothelioma, we examined whether the association between genital talc exposure and ovarian cancer risk is modified by variants of the NAT2 and GSTM1 genes, as well as the related glutathione S-transferase T1 (GSTT1) gene. The GSTM1 and GSTT1 genes produce enzymes involved in the metabolism of carcinogens and reactive oxygen species (13). These genes are homozygously deleted in approximately 50% (GSTM1) and 20% (GSTT1) of Caucasians, resulting in complete loss of enzymatic activity (14, 15). The NAT2 enzyme catalyzes the deactivation of xenobiotics via N-acetylation, but can also activate certain substrates via O-acetylation (16). Individuals with two NAT2 slow acetylator alleles, approximately 60% of individuals in Caucasian populations, have decreased rates of N- and O-acetylation (17-20). We hypothesized that the association between talc use and ovarian cancer risk would be stronger among individuals with the GSTM1 null, GSTT1 null, and NAT2 slow acetylator genotypes, due to decreased metabolism of free radicals and other products of the biologic response to talc. We examined these gene-talc interactions, as well as the main effect of talc use and each genotype, in two study populations with a total of 1,385 ovarian cancer cases.

METHODS

New England Case-Control Study

The New England Case-Control Study (NECC) consists of 1,231 epithelial ovarian cancer cases and 1,244 controls from Massachusetts and New Hampshire. Participants were enrolled in the study in two phases, from May 1992 to March 1997 (phase 1; 563 cases and 523 controls) or from July 1998 to July 2003 (phase 2; 668 cases and 721 controls). Participants completed a detailed questionnaire on potential risk factors for ovarian cancer and covariates of interest during an in-person interview with a trained interviewer. To avoid capturing changes related to disease status, interviewers asked participants about exposures that occurred at least one year prior to the date of diagnosis for cases or the interview date for controls. The institutional review boards of Brigham and Women's Hospital and Dartmouth Medical School approved both phases of the study, and all participants provided written informed consent.

During the two study phases, NECC researchers identified 2,347 incident cases of ovarian cancer through hospital tumor boards and state cancer registries; 1,845 (79%) of these cases were eligible, and 71% of the eligible cases were enrolled in the study. Study investigators identified potential controls using random digit dialing, drivers' license records, and Massachusetts town resident lists. Controls were frequency-matched to cases by age and state of residence. Of the potentially eligible controls contacted by investigators during phase 1, 68% were eligible and agreed to participate. During phase 2, 197 potential controls declined to be contacted by returning a postcard to “opt out” of the study; of the remaining potentially eligible controls who were contacted, 67% were eligible and enrolled in the study. The eligibility criteria and the reasons for non-enrollment of eligible cases are described elsewhere (21).

Over 95% of study participants provided a blood specimen at study enrollment. NECC researchers separated the heparinized blood samples into plasma, red blood cell, and buffy coat (white blood cell) components, extracted DNA from the buffy coat using Qiagen DNA extraction (Qiagen Inc., Valencia, CA), and stored the extracted DNA in freezers at a temperature of -80°C.

Nurses' Health Study

In 1976, 121,701 female registered nurses between the ages of 30 and 55 responded to a mailed questionnaire about known and suspected risk factors for disease, leading to the establishment of the Nurses' Health Study (NHS). Study participants completed follow-up questionnaires every two years, providing information on new diagnoses of disease and updated information on risk factors. Participation in the study has remained high throughout follow-up; between 1976 and 2004 the percentage of follow-up information obtained (questionnaire responses plus deaths) was 95.3%. The corresponding follow-up percentages for women who provided a white blood cell or cheek cell specimen were 98% and 99%, respectively. The Institutional Review Board of Brigham and Women's Hospital, Boston, MA approved both the NHS and this analysis, and all participants provided implied consent by completing and returning the baseline questionnaire.

In 1989 and 1990, 32,826 participants submitted a blood sample for use in genetic and other biomarker analyses. Details of the blood collection are described elsewhere (22). Between 2001 and 2004, 33,040 women without a blood specimen provided a buccal cell specimen. We used a mouthwash protocol to collect the buccal cell samples, based on evidence that this method provides slightly higher DNA yield and quality, compared with collection using a cytobrush (23). We extracted DNA from each specimen within one week of receipt using Qiagen DNA extraction (Qiagen Inc., Valencia, CA), and stored the DNA at -80°C.

NHS nested case-control study

We collected information on new diagnoses of ovarian cancer on each questionnaire, and we also obtained information on deaths due to ovarian cancer through family members, the National Death Index, and the U.S. Postal Service. We confirmed each diagnosis using methods described previously (24). For this analysis, we included all cases with a DNA specimen available from prior to diagnosis (incident cases), as well as cases who submitted a DNA specimen within four years after diagnosis (prevalent cases). We included the prevalent cases in the analysis due to the similarity of characteristics of these cases and the incident cases, and also because the interval of four years between diagnosis and DNA collection was less than the average survival time of 65.7 months for the incident cases. All cases were diagnosed prior to June 1, 2004 and had no history of a prior cancer, other than non-melanoma skin cancer.

We randomly selected three controls per case from the study participants who gave a buccal cell or blood specimen, who had not had a bilateral oophorectomy prior to the date of diagnosis of the matched case, and who had no history of cancer, other than non-melanoma skin cancer, as of the cycle of diagnosis of the case. We excluded 30 controls from the analysis due to unavailability of genotyping data (n=28) or because the participant was later diagnosed with ovarian cancer and was included in the analysis as a case (n=2). Cases and controls were matched on month and year of birth, DNA type, and menopausal status at diagnosis. For the blood collection, cases and controls were additionally matched on menopausal status and postmenopausal hormone (PMH) use status at blood draw, month/year and time of day of blood draw, and fasting status at blood draw, since these control selections were also used for analyses of plasma hormones and other biomarkers (25).

Exposure assessment

The phase 1 and 2 NECC questionnaires included multiple questions about regular use of talcum, baby or deodorizing powder as an adult. Specific questions asked about type of use (as a dusting powder to the genital area, sanitary napkins, underwear, or non-genital areas), frequency of use, age at first use, number of years used, and brand of powder used. The 1982 NHS questionnaire requested information on whether the participant had ever commonly applied talcum, baby, or deodorizing powder to the perineal area (no, <once/week, 1-6 times/week, or daily) or to sanitary napkins (yes/no). For this analysis, we defined regular genital talc use as application of powder to the genital/perineal region at least once per week. We also created a categorical variable for frequency of talc use, using the categories from the NHS questionnaire.

Genotyping methods

Genotyping was performed at the Dana Farber/Harvard Cancer Center High Throughput Genotyping Core (for the NAT2 polymorphisms and NHS GSTM1 and GSTT1 gene deletions) and the Molecular Epidemiology Research Laboratory at the Harvard School of Public Health (for the NECC GSTM1 and GSTT1 gene deletions). All samples were genotyped for three single nucleotide polymorphisms that identify the NAT2*5, NAT2*6, and NAT2*7 alleles. These alleles account for over 99% of slow acetylator alleles in Caucasian populations (16, 26). The NAT2 I114T (rs1801280), R197Q (rs1799930), and G286E (rs1799931) polymorphisms were genotyped using the 5' nuclease assay (Taqman) on the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA), in 384-well format. Individuals with two slow acetylator alleles were classified as NAT2 slow acetylators, while individuals with zero or one slow acetylator allele were classified as rapid acetylators.

The NECC samples were genotyped for the GSTM1 and GSTT1 gene deletions using multiplex polymerase chain reaction (PCR), and the PCR products were resolved on a 1.5% agarose gel. The NHS samples were genotyped for the two gene deletions using Taqman realtime PCR in 384-well format. For both the multiplex and real-time PCR assays, individuals were considered to have the GSTM1 or GSTT1 null genotype if no PCR product was present for the respective gene; all other individuals were classified as GSTM1 or GSTT1 present.

All DNA samples were whole genome amplified prior to genotyping. Laboratory personnel blinded to the case-control status of the samples performed all genotyping, and each plate included blinded replicate samples for quality control purposes. The replicate samples were 100% concordant for all genotypes except the NECC GSTM1 and GSTT1 gene deletions, which were 98% and 95% concordant respectively.

Statistical analysis

We used a chi-square test to examine whether the NAT2 polymorphisms were in Hardy-Weinberg equilibrium in each population, and also to examine the distribution of each genotype by case-control status. We conducted all analyses separately in the NHS and NECC populations using consistent exposure and covariate definitions and, after testing for heterogeneity in the results, pooled the estimates using a random effects model (27). We used conditional (NHS) and unconditional (NECC and NHS) logistic regression to model the multivariable-adjusted odds ratio (as an estimate of the relative risk [RR]) and 95% confidence interval (CI) for the main effect of genital talc use, the main effect of each gene, and each combined gene-talc variable. We tested for a linear trend with increasing frequency of talc use by using a continuous variable weighted by the midpoint of each frequency category, and we calculated the p-value for trend using the Wald test. To assess effect modification by genotype, we used unconditional logistic regression to model the association between talc use and ovarian cancer risk within each genotype stratum, and we calculated the p-value for interaction using the chi-square test for the difference between the log likelihoods for models with and without interaction terms between regular genital talc use and genotype. In addition to the analyses of total ovarian cancer, we examined associations with the serous invasive histologic subtype, based on evidence from prior studies that risk of this subtype may be more strongly associated with talc use.

We adjusted all analyses for the matching factors, duration of oral contraceptive use, parity, tubal ligation, body mass index (BMI), and duration of PMH use. Women with missing data for the continuous covariates were assigned the median value of the covariate for their study population. In the NHS, where covariate data are available from multiple questionnaire cycles, we used the data from two cycles (two to four years) prior to the cycle of diagnosis for each case and their matched controls, for consistency with the timeframe of the NECC covariate data. We examined additional covariates as potential confounders, including physical activity, smoking history, menopausal status, age at menopause, breastfeeding duration, and family history of ovarian or breast cancer, but did not include them in the final model because they did not substantially change our estimates. We performed all analyses using SAS version 9.1 (SAS Institute Inc., Cary, NC).

RESULTS

Our study population included 1,175 cases and 1,202 frequency-matched controls from the NECC and 210 cases and 600 matched controls from the NHS, for a total of 1,385 ovarian cancer cases and 1,802 controls. Of the NHS cases, 49 were prevalent and 161 were incident with respect to the time of DNA collection. Characteristics of the NHS prevalent and incident cases were generally similar, although a higher percentage of the prevalent cancers were endometrioid (20% vs. 9%) and a lower percentage were invasive (76% vs. 86%). In the NECC, 618 cases had serous histology (53%), 450 were serous invasive (38%), 153 were mucinous (13%), 172 were endometrioid (15%), and 232 had other/undifferentiated histology (20%). In the NHS, 111 cases were serous (53%), 93 were serous invasive (44%), 23 were mucinous (11%), 25 were endometrioid (12%), and 51 had other/poorly differentiated histology (24%).

Over 96% of the NECC participants and 98% of the NHS participants were of self-reported European ancestry. In analyses restricted to these participants, the results were similar to those for the entire study population; we therefore included all participants in our analyses to maximize our sample size. The distributions of ovarian cancer risk factors were similar in the NECC and NHS populations, although on average the NHS participants were older, had higher parity, and were more likely to have used PMH, in part due to differences in the NECC and NHS age distributions (Table 1). Within each study population the cases and controls differed with respect to the known risk factors for ovarian cancer. In addition, in the NECC the cases had higher mean BMI than the controls, and a larger percentage of the cases reported a history of genital talc use. The NHS prevalent and incident cases had similar BMI, tubal ligation history, duration of PMH use, duration of lactation, and genital talc use history; however, the prevalent cases were, on average, slightly younger (60 vs. 62 years), less likely to be postmenopausal (71% vs. 87%), and had lower parity (2.7 vs. 3.1 children), later age at menarche (13.1 vs. 12.5 years), and a longer mean duration of oral contraceptive use (60 vs. 41 months; results not shown).

Table 1.

Characteristics of ovarian cancer cases and controls in the New England Case-Control Study (NECC) and the Nurses' Health Study (NHS)

Characteristic	NECC			NHS‖
	Cases	Controls	P**	Cases	Controls	P**
N	1175	1202		210	600
Mean value (standard deviation)
Age in years*	51 (13)	51 (13)	0.37	62 (8)	62 (8)	0.93
Parity among parous women	2.5 (1.3)	2.8 (1.5)	<0.001	3.0 (1.3)	3.3 (1.5)	0.007
Duration oral contraceptive use (months)†	52 (54)	61 (55)	0.006	46 (42)	53 (49)	0.22
Body mass index (kg/m2)	26.3 (6.3)	25.7 (5.5)	0.02	25.7 (5.0)	25.7 (4.5)	0.94
Duration PMH use (months)†	78 (86)	74 (71)	0.64	96 (84)	85 (68)	0.18
Duration of lactation (months)‡	3.4 (8.6)	5.9 (12.2)	<0.001	2.9 (2.3)	3.6 (2.5)	0.002
Percent of study population
Parous	68	81	<0.001	89	93	0.05
Ever user of oral contraceptives	48	60	<0.001	42	45	0.57
History of tubal ligation	14	18	0.007	14	21	0.02
Ever user of PMH	17	20	0.14	71	63	0.02
Family history of ovarian cancer	5.1	2.8	0.004	9.1	3.7	0.002
Any history of genital talc use	29	24	0.003	40	39	0.79
Regular genital talc use (>=once/week)	27	20	<0.001	29	24	0.15
Daily genital talc use	16	12	0.006	18	13	0.08
Genotype frequencies, %
GSTM1 null	51	53	0.42	48	52	0.36
GSTT1 null	21	22	0.85	19	21	0.45
NAT2 slow acetylator§	63	64	0.74	59	67	0.05

^*Cases and controls in each study population were matched (NHS) or frequency-matched (NECC) on age

^†Duration of oral contraceptive use and postmenopausal hormone (PMH) use among ever users

^‡Total duration among parous women

^§NAT2 acetylation genotype based on analysis of three single nucleotide polymorphisms, I114T, R197Q, and G286E

^‖In the NHS, duration of lactation was collected in 1986, family history of ovarian cancer was first collected in 1992, and history of genital talc use was collected in 1982; for variables collected on multiple questionnaires, the value from two cycles (two to four years) prior to the date of diagnosis for each case was used for the case and their matched controls

^**P-values calculated using proc ttest (continuous variables) or a chi-square test (binary variables)

In the NECC, women with a history of regular genital talc use were older, had higher mean BMI, were less likely to have ever used oral contraceptives, were more likely to be postmenopausal, and were more likely to have used PMH (Table 2). Among parous women in the NHS, the mean age at first birth was lower for regular talc users. In addition, NHS participants who regularly used talc were less likely to have a history of smoking or tubal ligation. There was no difference in the genotype frequencies by genital talc use history in either study population.

Table 2.

Characteristics of participants in the New England Case-Control Study (NECC) and the Nurses' Health Study (NHS) by history of regular genital talc use (at least once per week)

Characteristic	NECC			NHS§
	No Regular Talc Use	Regular Talc Use	P‖	No Regular Talc Use	Regular Talc Use	P‖
Mean value (standard deviation)
Age in years	50 (13)	53 (12)	<0.001	61 (8)	62 (8)	0.64
Parity among parous women	2.7 (1.4)	2.7 (1.4)	0.64	3.2 (1.4)	3.3 (1.5)	0.40
Age at first birth among parous women	25.0 (5.1)	24.6 (4.9)	0.22	25.0 (3.5)	24.4 (3.0)	0.03
Duration oral contraceptive use (months)*	58 (55)	54 (54)	0.24	53 (49)	43 (42)	0.08
Body mass index (kg/m2)	25.7 (5.7)	27.0 (6.4)	<0.001	25.6 (4.6)	26.2 (4.8)	0.13
Duration PMH use (months)*	75 (74)	78 (86)	0.68	90 (74)	83 (70)	0.38
Duration of lactation (months)†	4.8 (10.8)	4.3 (10.3)	0.38	3.5 (2.5)	3.2 (2.4)	0.20
Physical activity (hours/week)	2.8 (5.0)	2.4 (3.8)	0.06	3.0 (2.3)	3.1 (2.4)	0.61
Percent of study population
Parous	74	75	0.75	93	92	0.81
Ever user of oral contraceptives	55	50	0.03	44	44	0.96
History of tubal ligation	16	16	0.97	22	13	0.008
Postmenopause	45	54	<0.001	81	81	0.86
Ever user of PMH	17	26	<0.001	66	64	0.73
Ever smoker	53	55	0.35	57	47	0.02
Family history of ovarian cancer	3.9	4.1	0.82	4.5	6.4	0.31
Genotype frequencies, %
GSTM1 null	52	52	0.72	51	49	0.66
GSTT1 null	21	22	0.59	22	17	0.15
NAT2 slow acetylator‡	63	66	0.23	65	63	0.58

^*Duration of oral contraceptive use and postmenopausal hormone (PMH) use among ever users

^†Total duration among parous women

^‡NAT2 acetylation genotype based on analysis of three single nucleotide polymorphisms, I114T, R197Q, and G286E

^§In the NHS, duration of lactation was collected in 1986, family history of ovarian cancer was first collected in 1992, and history of genital talc use was collected in 1982; for variables collected on multiple questionnaires, the value from two cycles (two to four years) prior to the date of diagnosis for each case was used for the case and their matched controls

^‖P-values calculated using proc ttest (continuous variables) or a chi-square test (binary variables)

All p-values for the tests for heterogeneity comparing the NECC and NHS results were greater than 0.05. Talc use was associated with increased risk of ovarian cancer in both study populations, although the confidence intervals were wide in the NHS due to the limited sample size (Table 3). In the pooled analysis, the relative risk for the association with regular genital talc use was 1.36 (95% CI=1.14-1.63) for total ovarian cancer and 1.60 (95% CI=1.26-2.02) for the serous invasive subtype. In addition, there were highly significant trends between increasing frequency of talc use and risk of both total and serous invasive ovarian cancer in the NECC (p-trend=0.002 for total and <0.001 for serous invasive ovarian cancer) and pooled analyses (p-trend<0.001 for both total and serous invasive ovarian cancer). Regular genital talc use was not significantly associated with risk of the endometrioid (RR=1.41, 95% CI=0.97-2.05) or mucinous (RR=1.28, 95% CI=0.85-1.92) histologic subtypes in the pooled analysis. In the NECC, use of talcum powder on non-genital body areas was unassociated with ovarian cancer risk (multivariable-adjusted RR, also adjusted for genital talc use=0.91, 95% CI=0.73-1.12).

Table 3.

Relative risks (RRs) and 95% confidence intervals (CIs) for the association between genital talc use and ovarian cancer risk in the New England Case-Control Study (NECC) and the Nurses' Health Study (NHS)

	NECC†			NHS†			Pooled‡
	Cases (%)	Ctrls (%)	RR (95% CI)	Cases (%)	Ctrls (%)	RR (95% CI)	RR (95% CI)
Total epithelial ovarian cancer:
N*	1175	1202		210	600
Regular genital talc use (>=once/week)
No	859 (73.2)	957 (79.7)	1.00 (ref.)	138 (70.8)	414 (76.0)	1.00 (ref.)	1.00 (ref.)
Yes	314 (26.8)	244 (20.3)	1.40 (1.15, 1.70)	57 (29.2)	131 (24.0)	1.24 (0.83, 1.83)	1.36 (1.14, 1.63)
Frequency of genital talc use
Never	832 (70.9)	916 (76.3)	1.00 (ref.)	120 (61.5)	352 (64.6)	1.00 (ref.)	1.00 (ref.)
<once/week	27 (2.3)	41 (3.4)	0.72 (0.43, 1.19)	18 (9.2)	62 (11.4)	0.98 (0.54, 1.79)	0.82 (0.55, 1.20)
1-6 times/week	123 (10.5)	96 (8.0)	1.33 (1.00, 1.79)	22 (11.3)	61 (11.2)	1.01 (0.57, 1.79)	1.26 (0.97, 1.63)
Daily	191 (16.3)	148 (12.3)	1.41 (1.10, 1.79)	35 (18.0)	70 (12.8)	1.44 (0.88, 2.37)	1.41 (1.14, 1.76)
P-trend§			0.002			0.18	<0.001
Serous invasive ovarian cancer:
N*	450	1202		93	263
Regular genital talc use (>=once/week)
No	310 (69.0)	957 (79.7)	1.00 (ref.)	60 (68.2)	177 (73.8)	1.00 (ref.)	1.00 (ref.)
Yes	139 (31.0)	244 (20.3)	1.62 (1.26, 2.09)	28 (31.8)	63 (26.3)	1.48 (0.82, 2.68)	1.60 (1.26, 2.02)
Frequency of genital talc use
Never	299 (66.6)	916 (76.3)	1.00 (ref.)	54 (61.4)	151 (62.9)	1.00 (ref.)	1.00 (ref.)
<once/week	11 (2.4)	41 (3.4)	0.65 (0.32, 1.33)	6 (6.8)	26 (10.8)	0.79 (0.29, 2.11)	0.70 (0.39, 1.24)
1-6 times/week	56 (12.5)	96 (8.0)	1.56 (1.08, 2.26)	12 (13.6)	25 (10.4)	1.64 (0.71, 3.79)	1.58 (1.12, 2.21)
Daily	83 (18.5)	148 (12.3)	1.61 (1.18, 2.20)	16 (18.2)	38 (15.8)	1.34 (0.65, 2.76)	1.56 (1.17, 2.08)
P-trend§			<0.001			0.29	<0.001

^*Frequencies do not add up to total N due to missing data on talc use

^†Unconditional (NECC) and conditional (NHS) logistic regression adjusted for age, study center (NECC only), duration of oral contraceptive use (months), parity (continuous), tubal ligation, body mass index (kg/m2, continuous), and duration of postmenopausal hormone use (months)

^‡P-values for tests for heterogeneity comparing the NECC and NHS results were all >0.38

^§Weighted by the midpoint of each category of genital talc use frequency and calculated using the Wald test

Among the controls in each population, the genotype frequencies for the NAT2 polymorphisms were in Hardy-Weinberg equilibrium and the distributions of the GSTM1 null, GSTT1 null, and NAT2 slow acetylator genotypes were consistent with previous reports of Caucasian populations (19, 28, 29). Comparing the prevalent and incident cases in the NHS, a nonsignificantly higher percentage of the prevalent cases were NAT2 slow acetylators (67% vs. 56%), but the GSTM1 and GSTT1 genotype distributions did not differ for the prevalent and incident cases (results not shown).

None of the genotypes examined were associated with ovarian cancer risk in the NECC or pooled analyses (Table 4). In the NHS, individuals with the NAT2 slow acetylator genotype had a significant 35% decrease in ovarian cancer risk (RR=0.65, 95% CI=0.45-0.95). The combined GSTM1 null/NAT2 slow acetylator and GSTT1 null/NAT2 slow acetylator genotypes were also inversely associated with risk in the NHS (RR=0.57, 95% CI=0.33-0.98 and RR=0.51, 95% CI=0.26-0.99, respectively), when compared with the GSTM1 or GSTT1 present, NAT2 rapid acetylator genotype. However, these associations were no longer statistically significant when pooled with the NECC estimates.

Table 4.

Relative risks (RRs) and 95% confidence intervals (CIs) for the association between GSTM1, GSTT1, and NAT2 genotype and epithelial ovarian cancer risk in the New England Case-Control Study (NECC) and the Nurses' Health Study (NHS)

	NECC†			NHS†			Pooled‡
	Cases (%)	Ctrls (%)	RR (95% CI)	Cases (%)	Ctrls (%)	RR (95% CI)	RR (95% CI)
N*	1175	1202		210	600
GSTM1 genotype
Present	573 (49.1)	567 (47.4)	1.00 (ref.)	102 (52.3)	268 (48.5)	1.00 (ref.)	1.00 (ref.)
Null	594 (50.9)	628 (52.6)	0.93 (0.79, 1.10)	93 (47.7)	285 (51.5)	0.83 (0.58, 1.17)	0.91 (0.78, 1.06)
GSTT1 genotype
Present	919 (78.8)	938 (78.5)	1.00 (ref.)	157 (81.3)	439 (78.8)	1.00 (ref.)	1.00 (ref.)
Null	247 (21.2)	257 (21.5)	0.98 (0.80, 1.21)	36 (18.7)	118 (21.2)	0.87 (0.57, 1.33)	0.96 (0.80, 1.16)
NAT2 genotype
Rapid/intermediate acetylator	387 (36.8)	405 (36.1)	1.00 (ref.)	77 (41.0)	182 (33.0)	1.00 (ref.)	1.00 (ref.)
Slow acetylator	665 (63.2)	717 (63.9)	0.97 (0.81, 1.15)	111 (59.0)	369 (67.0)	0.65 (0.45, 0.95)	0.82 (0.57, 1.20)
Combined GSTM1/GSTT1 genotype
Both present	445 (38.2)	430 (36.0)	1.00 (ref.)	81 (44.3)	206 (39.2)	1.00 (ref.)	1.00 (ref.)
M1 null, T1 present	474 (40.7)	508 (42.5)	0.91 (0.76, 1.10)	68 (37.2)	208 (39.5)	0.82 (0.54, 1.22)	0.89 (0.75, 1.06)
M1 present, T1 null	128 (11.0)	137 (11.5)	0.94 (0.71, 1.24)	17 (9.3)	49 (9.3)	0.98 (0.52, 1.84)	0.94 (0.73, 1.22)
Both null	119 (10.2)	120 (10.0)	0.94 (0.70, 1.26)	17 (9.3)	63 (12.0)	0.65 (0.34, 1.24)	0.88 (0.67, 1.15)
Combined GSTM1/NAT2 genotype
GSTM1 present, NAT2 rapid	195 (18.6)	188 (16.9)	1.00 (ref.)	37 (21.0)	85 (16.4)	1.00 (ref.)	1.00 (ref.)
GSTM1 null, NAT2 rapid	189 (18.1)	214 (19.2)	0.82 (0.61, 1.09)	35 (19.9)	91 (17.5)	0.96 (0.53, 1.74)	0.84 (0.65, 1.09)
GSTM1 present, NAT2 slow	315 (30.1)	343 (30.7)	0.86 (0.66, 1.11)	56 (31.8)	161 (31.0)	0.87 (0.51, 1.50)	0.86 (0.68, 1.09)
GSTM1 null, NAT2 slow	347 (33.2)	371 (33.2)	0.88 (0.68, 1.14)	48 (27.3)	183 (35.2)	0.57 (0.33, 0.98)	0.76 (0.50, 1.14)
Combined GSTT1/NAT2 genotype
GSTT1 present, NAT2 rapid	296 (28.3)	312 (28.0)	1.00 (ref.)	52 (29.7)	144 (27.5)	1.00 (ref.)	1.00 (ref.)
GSTT1 null, NAT2 rapid	88 (8.4)	90 (8.1)	1.03 (0.73, 1.45)	17 (9.7)	30 (5.7)	1.55 (0.76, 3.16)	1.11 (0.81, 1.52)
GSTT1 present, NAT2 slow	519 (49.7)	562 (50.4)	0.97 (0.79, 1.19)	90 (51.4)	270 (51.6)	0.87 (0.57, 1.33)	0.95 (0.79, 1.14)
GSTT1 null, NAT2 slow	142 (13.6)	152 (13.6)	0.98 (0.74, 1.31)	16 (9.1)	79 (15.1)	0.51 (0.26, 0.99)	0.76 (0.40, 1.43)

^*Frequencies do not add up to total N due to missing genotype data

^†Unconditional (NECC) and conditional (NHS) logistic regression adjusted for age, study center (NECC only), duration of oral contraceptive use (months), parity (continuous), tubal ligation, body mass index (kg/m2, continuous), and duration of postmenopausal hormone use (months)

^‡P-values for tests for heterogeneity comparing the NECC and NHS results were all >0.06

In analyses stratified by genotype, the association between regular genital talc use and risk of total ovarian cancer was stronger among women with the GSTT1 null and combined GSTM1 present/GSTT1 null genotypes (Table 5). In the pooled analysis, the relative risk for the association with regular genital talc use was 2.1 (95% CI=1.4-3.2) for women with the GSTT1 null genotype (p-interaction=0.03) and 2.8 (95% CI=1.6-5.0) for women with the GSTM1 present/GSTT1 null genotype (p-interaction=0.03). The association with the serous invasive subtype was also stronger within these genotype strata, although the p-values for interaction were not statistically significant. The pooled relative risk was 2.4 (95% CI=1.4-4.0) for the GSTT1 null stratum and 4.8 (95% CI=2.1-11) for the combined GSTM1 present/GSTT1 null stratum. The results were consistent in both study populations (results not shown), although the p-values for interaction were statistically significant only in the pooled analysis. There was also evidence of a stronger association between regular talc use and risk of serous invasive cancer among women with the GSTM1 present genotype, but this interaction was not statistically significant.

Table 5.

Pooled relative risks (RRs) and 95% confidence intervals (CIs) for the association between regular talc use and ovarian cancer risk, stratified by genotype, in the New England Case-Control Study (NECC) and the Nurses' Health Study (NHS)^*†

	All cancers		Serous invasive cancers		Cases and controls in pooled analysis
					All cases		Serous inv.		Controls
	Regular talc use		Regular talc use		Regular talc		Regular talc		Regular talc
	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes
Gene/stratum
GSTM1 genotype
Present (+)	1.0 (ref.)	1.6 (1.2, 2.0)	1.0 (ref.)	2.0 (1.4, 2.8)	480	189	173	90	646	165
Null (-)	1.0 (ref.)	1.3 (1.0, 1.6)	1.0 (ref.)	1.4 (1.0, 1.9)	498	179	190	76	690	198
P-interaction§	0.13		0.08
GSTT1 genotype
Present (+)	1.0 (ref.)	1.2 (1.0, 1.5)	1.0 (ref.)	1.5 (1.2, 2.0)	785	278	288	129	1035	301
Null (-)	1.0 (ref.)	2.1 (1.4, 3.2)	1.0 (ref.)	2.4 (1.4, 4.0)	194	87	71	38	300	67
P-interaction§	0.03		0.18
NAT2 genotype
Rapid/intermediate acetylator	1.0 (ref.)	1.5 (1.1, 2.0)	1.0 (ref.)	1.9 (1.2, 2.8)	330	128	123	57	459	113
Slow acetylator	1.0 (ref.)	1.4 (1.1, 1.8)	1.0 (ref.)	1.6 (1.2, 2.1)	552	217	204	96	819	233
P-interaction§	0.60		0.58
GSTM1/GSTT1 genotype‡
GSTM1+, GSTT1 +	1.0 (ref.)	1.4 (1.0, 1.8)	1.0 (ref.)	1.7 (1.2, 2.5)	378	142	136	68	479	140
GSTM1-, GSTT1+	1.0 (ref.)	1.2 (0.9, 1.5)	1.0 (ref.)	1.4 (0.9, 1.9)	400	135	151	60	541	154
GSTM1+, GSTT1-	1.0 (ref.)	2.8 (1.6, 5.0)	1.0 (ref.)	4.8 (2.1, 11)	98	47	34	22	158	24
GSTM1-, GSTT1-	1.0 (ref.)	1.6 (0.9, 2.9)	1.0 (ref.)	1.4 (0.6, 3.1)	94	40	36	16	138	41
P-interaction§	0.03		0.09

^*NECC: unconditional logistic regression adjusted for age, study center, duration of oral contraceptive use (months), parity (continuous), tubal ligation, body mass index (kg/m2, continuous), and duration of postmenopausal hormone use (months); NHS: unconditional logistic regression adjusted for age in months, menopausal status at diagnosis (post, pre/dubious), DNA source, duration of oral contraceptive use (months), parity (continuous), tubal ligation, body mass index (kg/m2, continuous), and duration of postmenopausal hormone use (months)

^†P-values for tests for heterogeneity comparing the NECC and NHS results were all >0.36

^‡NHS analysis adjusted for age, menopausal status at diagnosis, and DNA source only, to improve stability of estimates

^§P-values for interaction based on likelihood ratio test comparing unconditional logistic regression models with and without gene-talc interaction terms

We additionally analyzed the association between combined gene-talc variables, compared to a common referent group (wild-type genotype and no talc use), and risk of total and serous invasive ovarian cancer. The results of these analyses were similar to the stratified results presented in table 5, and are therefore included only as a supplementary table. We also examined interactions between regular genital talc use and combined GSTM1/NAT2 and GSTT1/NAT2 genotype (results not shown). The GSTT1 null/NAT2 slow acetylator genotype seemed to increase the risk of total and serous invasive ovarian cancer associated with talc use. However, these analyses were based on small numbers, especially for certain combinations of the genotype and talc variables, and none of the p-values for interaction were significant.

In analyses restricted to the NHS incident cases or the NHS cases and controls with a blood specimen, the results were similar to those for the total NHS study population (results not shown).

DISCUSSION

These results provide additional support for a main effect of genital talc exposure on risk of epithelial ovarian cancer. The presence of a significant trend between frequency of talc use and risk of total and serous invasive ovarian cancer in the NECC and pooled analyses further strengthens the evidence for an association, as most previous studies have not observed a dose-response with increasing frequency or duration of talc use (1, 5). The results of our gene-environment analyses suggest that genes in detoxification pathways may be involved in the biologic response to talc, and that the association between genital talc use and risk of ovarian cancer may vary by genotype. In particular, women with the GSTT1 null genotype and the combined GSTM1 present/GSTT1 null genotype had a stronger association between talc use and ovarian cancer risk. The evidence for these interactions was consistent in two independent study populations, and the p-values for interaction were statistically significant in a pooled analysis of the two populations. However, the direction of the interaction with combined GSTM1/GSTT1 genotype was unexpected based on the known function of these genes.

Although prior analyses of the talc/ovarian cancer association in the NHS and the NECC have been published, our study includes an additional 612 NECC cases and 679 NECC controls and eight additional years of follow-up in the NHS (3, 4). In the previous analysis of the NECC, Cramer et al. observed a significant positive association between talc use and risk of both total and serous invasive ovarian cancer. In addition, there was a significant trend with lifetime number of talc applications, after excluding applications during non-ovulatory intervals (p-trend=0.02), but no trend with duration or frequency of talc use (3). In the only prospective study of this association, Gertig et al. reported a significant association between talc use and risk of the serous invasive subtype in the NHS, but no association with risk of total ovarian cancer (4). Our findings are consistent with the previous reports for these study populations, although our analysis differs from the prior studies in that we defined our primary exposure variable as genital use of talc at least once per week, based on the assumption that habitual talc use is more likely to be recalled accurately and more likely to be associated with ovarian cancer risk. Our findings are also consistent with meta-analyses of this association (1, 30).

The controversy regarding the existence of an association between talc and ovarian cancer has stemmed in part from the lack of a clear mechanism for the association. Although talc and asbestos are chemically similar, their biologic effects may differ, since talc does not appear to be a lung carcinogen (31). In addition, it is unclear whether talc applied to the perineum can reach the ovaries, although some studies have demonstrated that inert particles can travel through the female genital tract to the fallopian tubes and ovaries (32, 33), and others have found talc particles in ovarian tissue (34-37). Recent studies have suggested additional potential mechanisms for an association between talc and ovarian cancer. Talc particles can induce an inflammatory response in vivo, which may be important in ovarian cancer risk (38). Normal ovarian cells treated with talc are more likely to undergo cell proliferation and neoplastic transformation, and cellular generation of reactive oxygen species increases with increasing exposure to talc (11). Recent studies by Cramer and colleagues also support the possibility of an immune-mediated mechanism for an association between talc and ovarian cancer and suggest that exposure of the lower genital tract to talc may be sufficient to cause changes, such as production of heat shock proteins, accumulation of talc in pelvic lymph nodes, or decreased levels of anti-MUC1 antibodies, that could increase ovarian cancer risk (39-41).

Although no prior studies have examined gene-talc interactions, the indication of a possible immune-related mechanism between talc and ovarian carcinogenesis and the evidence for gene-asbestos interactions suggest that genes involved in detoxification and inflammatory pathways could be important in the response to talc. Previous studies have indicated that NAT2 and GSTM1 genotype may modify the association between asbestos exposure and risk of malignant mesothelioma; however, not all studies have been consistent (7, 8, 42, 43), and for NAT2 the direction of the interaction differed in studies conducted in Finnish and Italian populations (7, 8, 42, 44). This suggests that interactions with these genes may be complex and might depend on additional factors, such as the presence of other gene variants, the type of asbestos, or the level of asbestos exposure (8).

The GSTM1 and GSTT1 genes produce enzymes that metabolize products of oxidative stress and catalyze the detoxification of carcinogens and other xenobiotics (45). The GSTM1 deletion and to a lesser extent the GSTT1 deletion may increase the risk of certain cancers; however, our study and previous analyses do not support a direct association between the GSTM1 or GSTT1 gene deletion and risk of ovarian cancer (13, 17, 28). While there is some overlap in GST substrate specificity, there are also differences in the substrates metabolized by the GSTM1 and GSTT1 enzymes, which could help to explain the opposite direction of the interactions we observed between talc use and GSTM1 and GSTT1 genotype (13, 17, 45). In studies of pleural malignant mesothelioma, the GSTM1 null genotype was associated with increased risk (7, 8, 42, 43) while the GSTT1 null genotype was unassociated with risk of malignant mesothelioma (8, 42, 43) but was associated with a significant decrease in risk of asbestosis in one study (46), providing support that some functions of the GSTM1 and GSTT1 enzymes may differ. The direction of the associations between the GSTM1 and GSTT1 deletions and risk of asbestos-related disease was opposite to the direction of the interactions with talc observed in our study; this could potentially be due to differences in the chemical structures of talc and asbestos or differences in the by-products produced during the biologic response to talc and asbestos. The NAT2 enzyme catalyzes the transfer of an acetyl group to its substrates, including carcinogens such as heterocyclic and aromatic amines, which can result in either activation or deactivation of these substances (17, 20). Approximately 60% of Caucasians have two NAT2 slow acetylator alleles and consequently have decreased rates of acetylation, which can either increase or decrease the risk of certain cancers depending on the substrate and the cancer site (17, 20). To our knowledge, no previous studies have examined the association between the NAT2 slow acetylator genotype and ovarian cancer risk. We did not observe strong evidence of a main effect of NAT2 genotype or an interaction between NAT2 genotype and talc exposure.

The novelty of this analysis and the assessment of gene-talc interactions in two independent study populations, one with a large number of cases and the other with prospective data on talc use and ovarian cancer incidence, are strengths of this study. However, although the pooled analysis included a large number of cases and controls, our power was still insufficient to detect interactions with certain combinations of genes and for specific histologic subtypes. In addition, while both study populations had extensive covariate data, the use of common exposure and covariate definitions resulted in the loss of some detail, particularly for the NECC. Information on talc use was only collected in 1982 in the NHS, so it is possible that some participants were misclassified with respect to their talc use history. However, the number of participants who began using talc after 1982, when the participants were between 36 and 61 years of age, is most likely small. Although we do not have data on age at initiation of talc use in the NHS, in the NECC approximately 95% of controls with a history of regular genital talc use reported first using talc before age 35. Recall or selection bias may have affected the results of the NECC analyses, due to the retrospective study design. However, the consistency of the NECC and NHS results suggests that biases related to study design were not a major problem, since, with the exception of the DNA for a subset of the cases, the NHS data were collected prospectively. In addition, the exposure definition of genital talc use at least once per week may have decreased the influence of recall bias in this analysis, since habitual talc use is likely to be recalled more accurately than sporadic use.

In summary, our findings suggest that variants of the GSTM1 and GSTT1 genes may modify the association between genital talc use and risk of total and serous invasive ovarian cancer. However, additional research is needed to confirm these findings and to explore potential mechanisms for these interactions, particularly for the stronger talc/ovarian cancer association among women with the GSTM1 present/GSTT1 null genotype. If confirmed, these findings would strengthen the evidence for the carcinogenicity of talc to the ovarian epithelium.