Recreational Physical Activity and Epithelial Ovarian Cancer: A Case-Control Study, Systematic Review, and Meta-analysis

There is evidence that physical activity has a protective effect against cancers of the colon and breast and possibly of the endometrium and prostate as well (1). It remains unclear, however, whether physical activity is associated with epithelial ovarian cancer risk. Hormonal factors are well known to be involved in the etiology of both breast (2) and endometrial cancer (3), and the protective effect of physical activity for these cancers is thought to be mediated through alterations in the levels of endogenous sex hormones (4). Although the causal pathway to ovarian cancer is unclear, hormonal factors have been implicated (5, 6) because high parity and oral contraceptive use are highly protective. We might therefore expect an analogous protective effect of higher levels of physical activity against the risk of ovarian cancer.

Physical activity may also influence ovarian cancer risk through a reduction in chronic inflammation (7), which has been hypothesized to play a role in ovarian carcinogenesis (8). In addition, regular high levels of exercise can cause anovulation (9, 10) and thereby potentially reduce ovarian cancer risk, which, according to the “incessant ovulation” hypothesis (11), is increased by the repeated injury to the ovarian epithelium from regular ovulation. Physical activity might also influence ovarian cancer risk through alterations in immune system functioning (12).

Although a number of studies have examined the relationship between physical activity and ovarian cancer, their findings are inconsistent. This may be due, in part, to the use of different definitions of physical activity, different parameters of activity (type, frequency, duration, intensity), and different methods of measurement. In addition, although evidence is emerging that the different histologic subtypes of ovarian cancer have different risk factor profiles (13, 14), only a small number of studies have examined the effects of physical activity separately for the subtypes (15-19), and only one of these studies examined the effects for all four major subtypes (16). In addition, there is no consensus on the critical exposure period during which physical activity may have the greatest effect on ovarian cancer risk.

We have endeavored to resolve some of these uncertainties by examining the influence of recreational physical activity on the risk of the different subtypes of epithelial ovarian cancer in a large national case-control study conducted in Australia. We have also included our data in a systematic review and meta-analysis of all available evidence linking recreational physical activity with epithelial ovarian cancer to obtain the best summary estimate of the association.

Study Participants. The Australian Ovarian Cancer Study was an Australia-wide population-based case-control study of epithelial ovarian cancer. The study methods have been described elsewhere (20). Cases were women ages 18 to 79 years, living in Australia with newly diagnosed histologically confirmed epithelial ovarian, fallopian tube, or primary peritoneal cancer in the period between January 2002 and June 2005. Cases were recruited by trained nurses who liaised with the treatment clinics, physicians, and state cancer registries throughout Australia.

Of 3,553 women identified with suspected ovarian cancer (most women were approached before surgery and thus before histologic diagnosis), 304 died before contact could be made, physicians refused to give consent to contact 133 usually because they were too sick or unable to give informed consent, and 194 women could not be contacted. A further 167 (5%) were excluded on the basis of language difficulties (n = 70), mental incapacity (n = 33), and illness (n = 64). The remaining 2,755 women were invited to participate, and of these, 2,319 (84%) agreed to take part. Two researchers independently abstracted information on tumor site, histologic subtype, and tumor behavior (invasive versus borderline) from the diagnostic histopathology reports, and discrepancies were resolved by consensus. For a sample of 87 women, the pathology reports and full set of diagnostic slides were reviewed by a gynecologic pathologist and the agreement with the original abstracted data was >97% for tumor site, behavior, and subtype. Following the histopathology review, 624 women were excluded because their final diagnosis was not confirmed as epithelial ovarian cancer, and 10 because their cancer was diagnosed before the start of the study period. Of the final 1,685 eligible participants, 1,580 (94%) returned a questionnaire.

Controls were randomly selected from the national electoral roll and were frequency matched by age (in 5-year age bands) and state of residence to the case group. Selected women were mailed an invitation letter and information brochure explaining the study and then, where possible, followed up by telephone. At least five attempts were made to reach each woman; those who did not have a listed telephone number were mailed a second invitation letter. Of the 3,600 women contacted and invited to participate, 158 were unable to complete the questionnaire due to illness (n = 61) or language difficulties (n = 97). Of the 3,442 remaining women, 1,612 (47%) agreed to participate and returned a completed questionnaire. Six of them reported a history of ovarian cancer and 97 reported a previous bilateral oophorectomy and thus were excluded from the present study, leaving 1,509 population controls.

The study was approved by the ethics committees of the Queensland Institute of Medical Research, Peter MacCallum Cancer Centre, and all participating hospitals and cancer registries.

Information was collected using a self-administered questionnaire, which included questions about demographic, medical, hormonal, reproductive, diet, family history, and other potential risk factors. Exposures were assessed before a reference date, defined as 1 year before the date of diagnosis (or date of first approach for controls), because it was felt that more recent exposures in cases could be influenced by the presence of subclinical disease. Recreational physical activity was assessed using two questions. First, women were asked, for different age periods (ages 10-19, 20-29, 30-49, and 50-79 years), how often they engaged in strenuous physical activity (i.e., “activity that made you puff heavily, e.g., running, aerobics, and sports at a competitive level such as tennis, swimming laps, squash”) for at least 20 min in their leisure time (never, less than once a month, less than once a week, once a week, 2-3 times per week, 4+ times per week). They were asked the same question about moderate physical activity (e.g., brisk walking, heavy gardening, sports at a social level such as tennis, golf, and cricket). These questions incorporate the “frequency” and “intensity” components of physical activity measured by the “Historical Leisure Activity Questionnaire” developed and validated by Kriska (21) and were validated in the Iowa Women's Health Study Cohort (22, 23). We examined responses to these two questions individually, and for comparability with other studies, we also combined them to form a three-level physical activity index (PAI; low, moderate, and high) based on frequency and intensity of activity (23). The “high” category included women who participated in strenuous activity 2+ times per week or moderate activity 4+ times per week. The “medium” category included women who participated in strenuous activity once per week or moderate activity one to three times per week. The “low” category comprised the remaining women (participating in strenuous or moderate activity less than once per week).

Physical activity data were incomplete for 99 cases and 35 controls who self-completed an abbreviated form of the questionnaire, which did not include questions about lifetime physical activity. These women were excluded from the analyses.

Unconditional logistic regression was used to estimate the odds ratios (OR) and 95% confidence intervals (95% CI) for the association between physical activity and ovarian cancer risk. Multivariable logistic models were used to adjust for potential confounders, including age at diagnosis/first approach (in 10 year age-groups as this gave the most efficient control for confounding by age), education (secondary school, technical college/apprenticeship, university), parity (0, 1-2, ≥3 full-term births), hormonal contraceptive use (none, <60 months, ≥60 months), and body mass index (BMI). BMI was classified using the WHO definitions of obesity: <18.5 kg/m², “underweight”; 18.5 to 24.9 kg/m², “normal weight”; 25 to 29.9 kg/m², “overweight”; and ≥30 kg/m^2, “obese” (24). Other potential confounders that were considered for all analyses but not included in the final models because they did not substantially alter risk estimates were smoking (current, former, never), total energy intake (quartiles), perineal talc use, history of hysterectomy or tubal sterilization, menopausal status, use of hormone replacement therapy, family history of breast or ovarian cancer in a first-degree relative, breast-feeding, and state of residence. Tests for linear trend were done using the maximum likelihood test with the categorical variable of interest entered as a continuous term.

We conducted analyses for invasive and borderline tumors first for all histologic subtypes combined, and then separately by subtype. We also examined the association with recent physical activity stratified by recent BMI and that with activity at age 20 to 29 years stratified by BMI at age 20 years (using WHO-defined categories of BMI). We conducted subgroup analysis to assess the interaction between menopausal status and all measures of recent physical activity (moderate, strenuous PAI). We defined “recent” physical activity as that occurring in the time period that included the woman's age at the reference date. The statistical significance of any observed stratum-specific differences was assessed by including a cross-product term in regression models. We classified women as postmenopausal if they reported natural or medical menopause before the reference date. Women for whom menopausal status could not be determined were excluded from this stratified analysis. All statistical analyses were done using SAS version 9.1 (SAS Institute, Inc.).

Selection of Studies. Eligible studies were identified using PubMed software to search Medline (U.S. National Library of Medicine, Bethesda, MD) for relevant articles from 1950 to February 2007 and by hand-searching the reference lists of the retrieved articles. For computer searches, we used the following MeSH terms or text words: “physical activity”, “exercise”, “activity”, or “recreational physical activity”, combined with “ovarian cancer”, “ovarian malignancy”, or “ovarian neoplasm”. Studies that had been commonly cited in the literature were also included as citation search terms in the ISI Science Citation Index (1990-present) to identify other studies that had referenced them. The search was not limited to studies published in English.

We read the abstracts of all identified studies to exclude those that were clearly not relevant. The full texts of the remaining articles were read to determine if they met the study inclusion criteria. Where multiple reports from one study were found, the most recent or most complete publication was used. Studies were included in the systematic review if they were a case-control or cohort study that permitted assessment of the association between physical activity (recreational/occupational/sedentary behavior) and histologically confirmed ovarian cancer. Studies that permitted quantitative assessment of an association between recreational physical activity and ovarian cancer were included in the meta-analysis.

The following information was recorded for each study: study type, years of data collection (case-control studies), duration of follow-up (cohort studies), age range of participants, country, variables for which statistical adjustment was done, number of cases and controls or person years, definitions and categories of physical activity exposures, point estimates [relative risk (RR), OR, or standardized incidence ratio], and 95% CIs. We included studies reporting the different measures of RR because ovarian cancer is a rare disease, and in such instances ORs and standardized incidence ratios provide a valid estimate of the RR. Where several risk estimates were presented, we used those adjusted for the greatest number of potential confounders. Where risk estimates for physical activity at different time periods were presented, we used the most recent period for case-control studies and either baseline measurements or cumulative averages, if available, for the cohort studies.

To pool RR estimates for the highest category of recreational physical activity compared with the lowest, a weighted average of the log RR was estimated, taking into account the random effects using the method of DerSimonian and Laird (25). We assessed heterogeneity for each pooled estimate with a Cochran's Q test for heterogeneity. We also conducted separate analyses by study type. Finally, we conducted sensitivity analyses, omitting each study in turn to determine whether the results could have been influenced excessively by a single study. We evaluated publication bias by qualitatively assessing a funnel plot of the natural logarithms of the effect estimates for the risk of ovarian cancer related to recreational physical activity against their variance (26).

Based on the histopathology review, 1,269 women had invasive cancer classified as follows: serous, 846 (67%); endometrioid, 138 (11%); clear cell, 88 (7%); mucinous, 44 (3%); and mixed histopathology, 153 (12%). A further 311 women had borderline (low malignant potential) tumors classified as follows: serous, 152 (49%); mucinous, 147 (47%); endometrioid, 3 (1%); and mixed, 9 (3%). This distribution is consistent with previous studies (14, 27, 28). Cases were significantly older than controls (mean age: cases, 57.9 years; controls, 56.4 years; P = 0.001) and were less likely to have continued their education beyond high school. As expected, cases were more likely to be nulliparous and to report a history of breast or ovarian cancer in a first-degree relative, but were less likely to have ever used oral contraceptives (Table 1).

Table 2 presents the ORs for invasive and borderline ovarian cancer associated with recent recreational activity, both moderate and strenuous, together with recent PAI and PAI at different ages. Compared with women in the lowest categories of recent moderate or strenuous physical activity and PAI, the multivariable-adjusted ORs (and 95% CIs) for the highest category of activity were 0.8 (0.6-1.1) for moderate activity, 1.0 (0.8-1.4) for strenuous activity, and 0.9 (0.7-1.1) for PAI for invasive tumors. Inclusion of both strenuous and moderate activity in the same model made no material difference to the results. Results were similar for borderline tumors. Similarly, associations with PAI at different ages were mostly weakly inverse or null.

Overall, the relationship between recreational physical activity and risk of ovarian cancer did not differ substantially for the different histologic subtypes, although moderate levels of recent PAI were associated with a nonsignificant increased risk of mucinous borderline tumors (OR, 1.4; 95% CI, 0.8-2.3) and a reduced risk of invasive endometrioid tumors (OR, 0.6; 95% CI, 0.4-1.1; Table 3). The relationship of recent physical activity (moderate, strenuous, PAI) and physical activity at age 20 to 29 years (PAI) with ovarian cancer risk did not differ materially for women in different WHO categories of BMI (data not presented). We found no effect modification by menopausal status.

The primary computerized literature search identified 129 potentially relevant studies. After review of the study abstracts, we retrieved 20 articles for further assessment, of which 7 reports of cohort studies (18, 22, 27, 29-32) and 9 from case-control studies (15-17, 33-39) met the criteria for inclusion in the systematic review. Of these, 12 studies were restricted to epithelial ovarian cancer and the other 4 included predominantly epithelial ovarian cancers.

Twelve of the 16 studies assessed recreational physical activity [six case-control studies (15-17, 34-36) and six cohort studies (18, 22, 27, 29, 30, 32)] and all presented risk estimates permitting meta-analysis together with the results of the current study. These studies are summarized in Table 4. One of the cohort studies (32) used “moderate” rather than “low” as the reference category; however, there was very little difference in risk between the low and moderate categories in this study, and thus the OR for high versus moderate was included in the meta-analysis.

For all studies, the pooled RR of ovarian cancer in women in the highest category of recreational physical activity compared with those in the lowest category was 0.81 (95% CI, 0.72-0.92) with significant heterogeneity (P = 0.03; Fig. 1). After stratifying by study design, the pooled RRs were 0.79 (95% CI, 0.70-0.85) for case-control studies and 0.81 (95% CI, 0.57-1.17) for cohort studies. Heterogeneity was evident among the cohort studies (P = 0.004) due almost entirely to the outlying result of Anderson et al. (22). Excluding this study gave a pooled RR of 0.72 (95% CI, 0.53-0.98) with no significant heterogeneity (P = 0.14). We observed no heterogeneity among the case-control studies (P = 0.70), with pooled RRs ranging from 0.71 to 0.84 after omitting individual studies. All studies adjusted for age and parity and most also adjusted for oral contraceptive use and BMI. Excluding studies that did not adjust for oral contraceptive use (15, 22, 30, 34) and BMI (18, 22, 30, 32) resulted in pooled estimates of 0.83 (95% CI, 0.80-0.86) and 0.81 (95% CI, 0.76-0.86), respectively, with no significant heterogeneity. The funnel plot of the effect estimates for the risk of ovarian cancer related to recreational physical activity was close to symmetrical, suggesting that there was no appreciable publication bias.

Several case-control (15, 17, 34, 36) and cohort studies (18, 27, 29, 31, 32), as well as the current study, assessed the effect of recreational physical activity stratified by BMI. Two case-control studies (15, 36) observed larger risk reductions with higher levels of activity for obese women. In contrast, all other studies including our own found no significant effect modification of the relationship between physical activity and ovarian cancer risk by BMI.

Four cohort studies (18, 22, 30, 32) and three additional case-control studies (15, 34, 37) reported on “vigorous” recreational activity and risk of ovarian cancer. Two of these, one case-control (37) and one cohort (30), reported significant inverse associations with vigorous physical activity. One cohort study (22) reported a significant positive association, whereas the other four studies, like ours, found no significant association (15, 18, 32, 34). One of the case-control studies and one additional study also found no significant relationship with vigorous occupational activity (37) or total (recreational and occupational) vigorous activity (31).

Two case-control studies (33, 38) and one cohort study (27) reported the results of analyses of sedentary behavior, based on duration of sitting, and risk of ovarian cancer. Patel et al. (27) reported a significantly increased risk for the highest category of hours of sitting per day and Zhang et al. (33) reported a nonsignificant increased risk, whereas Dosemeci et al. (38) reported a nonsignificant inverse association.

Five case-control studies (15, 17, 37-39) and one cohort study (27) presented results for nonrecreational or occupational physical activity. Whereas the cohort study (27) and two case-control studies (15, 37) found no relationship with risk of ovarian cancer, the other three reported modest inverse associations with higher levels of occupational physical activity (17, 38, 39).

A further five case-control (15-17, 34, 35) and three cohort studies (18, 27, 32) have examined whether there are critical exposure periods in which physical activity may influence ovarian cancer risk. Although caution must be taken in interpreting these studies due to the use of different age-categories/time periods of measurement, these studies did not find differential effects of physical activity on ovarian cancer risk across time, showing either protective (15-17, 32, 35) or null effects (18, 27, 34) across all time/age periods.

The current study, three other case-control studies (15-17) and two cohort studies (18, 19) have reported on the effects of physical activity separately for the different histologic subtypes of ovarian cancer. Generally, there were no notable differences between the subtypes, although there was a suggestion in two previous case-control studies and the present study that effects might differ for mucinous tumors (15, 16). Pan et al. (15) observed inverse associations for moderate activity for all subtypes except mucinous; they also observed an increased risk of mucinous tumors associated with high levels of vigorous activity. Riman et al. (16) found protective effects for recreational physical activity for all subtypes during all ages except for activity during the ages of 18 to 30 years and mucinous tumors. The present study observed a nonsignificant increased risk of mucinous borderline tumors associated with medium levels of recreational physical activity (Table 3).

Overall, the findings of our population-based case-control study point to a null or, at most, weakly inverse association between recreational physical activity and risk of epithelial ovarian cancer. Our meta-analysis, however, shows a consistent and significant pattern of weak to modest inverse associations between recreational physical activity and epithelial ovarian cancer risk despite the variation in populations, study design, and adjustment in analyses. Heterogeneity among studies was due to a single cohort study (22); this was similar in design and conduct to the other cohort studies, although it lacked adjustment for oral contraceptive use or BMI. It was also apparent graphically (Fig. 1) that a number of studies suffered from a lack of power to detect modest effects and were unable to exclude chance as an explanation for their findings when considered in isolation. These studies, however, revealed a consistent pattern of decreased risk associated with higher levels of recreational physical activity, leading to a summary effect estimate of 0.81 (95% CI, 0.72-0.92). These observations underline the importance of considering new data carefully in the context of all previous findings whenever possible. Overall, the evidence was less consistent for occupational activity, vigorous activity, and sedentary behavior, and fewer studies had examined these measures. Although it has been suggested that there may be critical time periods throughout a woman's life where physical activity may have the most effect on ovarian cancer risk (18), our study and the findings of others do not point to a differential effect across time.

Possible limitations of the meta-analysis include potential residual confounding of the associations due to different degrees of control for confounding variables in different studies. All studies, however, adjusted for important confounding factors such as age and parity, and most also adjusted for oral contraceptive use and BMI. Given that there was no material difference in the summary estimate when the studies that did not control for these latter characteristics were excluded, we are confident that the observed relationship between recreational physical activity and risk of ovarian cancer was not influenced by these factors. Second, the studies contributing to the summary estimates were vulnerable to various types of bias. The majority relied on self-reports of physical activity, which are prone to overestimation, particularly by overweight and obese subjects (40). This could have resulted in nondifferential misclassification, which may have led to underestimation of the true association between recreational physical activity and ovarian cancer. Differential misclassification was also a possibility in the case-control studies, but this is unlikely because cases would not readily associate their levels of physical activity with their cancer. Selection bias due to self-selection of more health-conscious control women, who are more likely to be physically active, also needs to be considered, but the consistency of results from the case-control and cohort studies suggests that such a bias is unlikely. Third, publication bias is always possible, and we searched only indexed journals. The funnel plot of the effect estimates for the risk of ovarian cancer related to recreational activity, however, was close to symmetrical, suggesting that there was no appreciable publication bias, and in general positive findings are more likely to be reported than otherwise (41). Thus, on balance, we believe that our systematic review provides a reasonably valid summary of the available evidence.

Strengths of the current study include the population-based design, large number of cases, and detailed information on multiple exposures. A limitation was the relatively low participation rate among controls (47%); however, the consistency of our findings with previous research on physical activity and ovarian cancer suggests that it is unlikely that nonresponse could have biased the results appreciably. A comparison with statistics from the Australian National Health Survey conducted in 2004 (a representative survey of the Australian adult population) revealed that the distributions of education level, parity, and BMI among our control women were almost identical to those from the National Health Survey;³

There are several biological mechanisms whereby physical activity could reduce the risk of ovarian cancer, and these include alterations in endogenous sex hormone levels (estrogen, progesterone, and androgens), suppression of ovulation, insulin-mediated pathways, and maintenance of energy balance (1). Regular physical activity has been shown to lower the levels of biologically available estrogens, progesterone, and androgens (42-46) and increase levels of circulating sex hormone binding protein (47). There is a significant body of evidence suggesting that androgens may increase ovarian cancer risk whereas progesterone plays a protective role (6). Regular strenuous physical activity can increase the probability of anovulation and amenorrhea (9), which may offer protection according to the “incessant ovulation” hypothesis (10). Regular physical activity also significantly lowers insulin levels and enhances insulin sensitivity, independently of BMI (48-50), and insulin increases the bioactivity of insulin-like growth factor I (51, 52). Lukanova et al. (53) found a strong direct relationship between circulating insulin-like growth factor I levels and risk of developing ovarian cancer before age 55 years, and high levels of insulin-like growth factor I have been associated with other hormone-related cancers: prostate and breast cancers (54, 55).

Regular physical activity also helps prevent or reduce obesity with consequent improvement in the metabolic profile (endogenous hormone and growth factor levels) (1). A recent meta-analysis concluded that obesity is a modest but significant risk factor for ovarian cancer (56). Physical activity may also influence carcinogenesis through alterations in immune system functioning (12), a reduction in chronic inflammation (7, 8), or other associated lifestyle factors (57).

In summary, our meta-analysis points to a consistent, albeit weak, inverse relationship between recreational physical activity and the occurrence of epithelial ovarian cancer. The cohort data indicate that the relation does indeed refer to activity that precedes the origins of the cancers, and there are a number of biologically plausible pathways that may underlie a protective effect of higher levels of activity. There are also analogies with stronger findings for other hormonal tumors. Hence, it is reasonable to speculate that the observed effect could be causal. In contrast to many known risk factors for ovarian cancer, physical activity is potentially a more modifiable behavior. This therefore potentially offers women an opportunity to reduce their risk of ovarian cancer in addition to the numerous other chronic diseases associated with low physical activity.

We thank the New South Wales, Queensland, South Australian, Victorian and Western Australian Cancer Registries as well as all the collaborating institutions represented within the Australian Ovarian Cancer Study Group (below). We also thank all of the women who participated in the study.

The Australian Ovarian Cancer Study Group. Management Group: D. Bowtell [Peter MacCallum Cancer Centre (PMCC)], G. Chenevix-Trench, A. Green, P. Webb [Queensland Institute of Medical Research (QIMR)], A. deFazio (Westmead Hospital), D. Gertig (University of Melbourne). Project Managers: N. Traficante (PMCC), S. Moore (QIMR), J. Hung (Westmead Hospital). Data Managers: S. Fereday (PMCC), K. Harrap, T. Sadkowsky (QIMR). Research Nurses: New South Wales—A. Mellon, R. Robertson (John Hunter Hospital), T. Vanden Bergh (Royal Hospital for Women), J. Maidens (Royal North Shore Hospital), K. Nattress (Royal Prince Alfred Hospital), Y.E. Chiew, A. Stenlake, H. Sullivan (Westmead Hospital); Queensland—B. Alexander, P. Ashover, S. Brown, T. Corrish, L. Green, L. Jackman, K. Martin, B. Ranieri (QIMR); South Australia—J. White (QIMR); Tasmania—V. Jayde (Royal Hobart Hospital); Victoria—L. Bowes (PMCC), P. Mamers (Monash Medical Centre); Western Australia—T. Schmidt, H. Shirley, S. Viduka, H. Tran, S. Bilic, L. Glavinas [Western Australia Research Tissue Network (WARTN)]. Clinical Collaborators: New South Wales—A. Proietto, S. Braye, G. Otton (John Hunter Hospital); T. Bonaventura, J. Stewart (Newcastle Mater Misericordiae); M. Friedlander (Prince of Wales Hospital); D. Bell, S. Baron-Hay, A. Ferrier, G. Gard, D. Nevell, B. Young (until mid-2003; Royal North Shore Hospital); C. Camaris, R. Crouch, L. Edwards, N. Hacker, D. Marsden, G. Robertson (Royal Hospital for Women); P. Beale, J. Beith, J. Carter, C. Dalrymple, A. Hamilton, R. Houghton, P. Russell (Royal Prince Alfred Hospital); A. Brand, R. Jaworski, P. Harnett, G. Wain (Westmead Hospital); Queensland—A. Crandon, M. Cummings, K. Horwood. A. Obermair, D. Wyld [Royal Brisbane and Women's Hospital (RBWH)]; J. Nicklin (RBWH and Wesley Hospital), L. Perrin (RBWH and Mater Misericordiae Hospitals), B. Ward (Mater Misericordiae Hospitals); South Australia—M. Davy, C. Hall, T. Dodd, T. Healy, K. Pittman (Royal Adelaide Hospital, Burnside Memorial Hospital); D. Henderson, S. Hyde (Flinders Medical Centre); J. Miller, J. Pierdes (Queen Elizabeth Hospital); Tasmania—P. Blomfield, D. Challis, R. McIntosh, A. Parker (Royal Hobart Hospital); Victoria—B. Brown, R. Rome (Freemasons Hospital); D. Allen, P. Grant, S. Hyde, R. Laurie, M. Robbie (Mercy Hospital for Women); D. Healy, T. Jobling, T. Maniolitas, J. McNealage, P. Rogers, B. Susil, A. Veitch, J. Constable, S. Ping Tong, I. Robinson, I. Simpson (Monash Medical Centre); K. Phillips, D. Rischin, P. Waring, M. Loughrey, N. O'Callaghan, Bill Murray (PMCC); V. Billson, S. Galloway, J. Pyman, M. Quinn (Royal Women's Hospital); Western Australia—I. Hammond, A. McCartney, Y. Leung (King Edward Memorial Hospital, St John of God). Scientific Collaborators: I. Haviv (PMCC); D. Purdie, D. Whiteman (QIMR); N. Zeps (WARTN).

The Australian Cancer Study Group investigators are A.C. Green, P.G. Parsons, N. Hayward, P. Webb, D. Purdie, and D. Whiteman (QIMR).