Skip to main content
         

Vol. 11 Issue 1, Winter 2006

Research Commentary: Induced Abortion is Not Associated with Breast Cancer Risk
The Ribbon 

Barbour S. Warren, Ph.D.
Research Associate, BCERF Sprecher Institute for Comparative Cancer Research

The recent Supreme Court appointments have made everyone aware once again of the high degree of contention that exists surrounding the issue of abortion. This contention has carried over into the area of breast cancer risk. Several groups are claiming that abortion is related to an increase in breast cancer risk. Currently, three states, Mississippi, Missouri and Texas, are mandating pre-abortion counseling discussing breast cancer risk (Guttmacher Institute, 2006). In February 2003, the National Cancer Institute held a workshop with scientific experts in this area of research and updated their position on this issue (National Cancer Institute, 2003). The summary statement from this workshop conveyed that it is "well established" that induced abortion is not associated with breast cancer risk. This was a strong statement as "well established" is the highest category for their strength of scientific evidence evaluation. Similar statements have also been made by the World Health Organization (World Health Organization, 2000), the American Cancer Society (American Cancer Society, 2005) and both the American and Royal (British) Colleges of Obstetricians and Gynecologists (ACOG, 2003; RCOG, 2004). This article will review this area of research by examining the issues that were likely to have been important in these evaluations. An earlier article in The Ribbon presented the decisive types of scientific evidence that are necessary to define a cause-effect relationship. This article will examine whether the scientific evidence for a relationship between induced abortion and breast cancer meets these standards. Although there are more than sixty studies that have examined this issue, most of the examples in this discussion will be drawn from a collaborative reanalysis in which the researchers in almost all of the existing studies concerning abortion and breast cancer shared all their data, which was pooled and reanalyzed as a single set (Beral et al., 2004). This method allowed for compensation for many of the shortcomings of the individual studies.

First, a little background on epidemiological studies is in order. There are two types of epidemiological studies that are relevant in this evaluation. The first type is known as a cohort study. In this type of study, information is gathered for a large group of women without breast cancer. Following a period of time, the information from the cohort women who have gotten breast cancer and those who have not is compared and analyzed. Cohort studies are considered to be most reliable as healthy people are thought to have less bias and provide more accurate information. The second type of study is a case-control study. In this type of study, information about abortion and breast cancer risk factors is gathered and analyzed for women who have breast cancer and similar women who do not have breast cancer. The majority of studies that have examined abortion and breast cancer risk have been case-control studies.

Study characteristics which contributed to varying results

As a bit more background, there are several problems that can occur when these studies are carried out.

Reference Group Choice. The first problem is connected with the group that is chosen as the reference group for risk comparisons. Epidemiological studies define risk factors by comparing the incidence of breast cancer in an "exposed group" (in this case women who have had abortions) to the incidence of breast cancer in a similar but "unexposed group" (women who have not had abortions). Giving birth causes changes in the breast that reduce breast cancer risk and a study's results would be different depending on whether the reference "unexposed" group was made up of women who have given or not given birth. Some studies have compared the breast cancer risk of women who had abortions to women who had not given birth, others to women who had given birth, and some studies did not designate. These three study designs would be expected to produce different levels of breast cancer risk.

Reporting Bias. The second study design problem involves the accuracy of the determination of whether women in the study group have had abortions; this is known as reporting bias. Study participants typically do not know what aspect of breast cancer risk is under examination and women with breast cancer have been found to more freely report having had an abortion than women without breast cancer. The resulting skew of information can have a substantial effect on case-control studies examining the association of induced abortion and breast cancer risk.

Adjustment for "Confounding." Careful epidemiological studies account for the effect of members of the study groups having established breast cancer risk factors. For example, age is a substantial risk factor and almost all studies correct for differences in age between the "exposed" and "unexposed" groups under study. This is known as adjusting for confounding. Many of the abortion and breast cancer studies have adjusted for only a few breast cancer risk factors and their results are less trustworthy.

Effect Modification. There are other characteristics that might affect the change in the strength of association between abortion and breast cancer risk. Such things as the length of the aborted pregnancy, the women's reproductive history, and the number of abortions could potentially have effects. These aspects are known as effect modifiers and their contribution to any resulting breast cancer risk may or may not have been accounted for.

Major standards for demonstrating cause-effect relationships

Now, on to how well the scientific evidence supports the idea that abortion has a cause-effect relationship with breast cancer; i.e. how well the evidence fits the standards for a cause-effect relationship.

No consistent association across studies. The first of the cause-effect standards is that the available studies show a consistent association between induced abortion and breast cancer. It would be expected that if a relationship existed that most studies would find and report it consistently. Studies of abortion and breast cancer risk have not met this standard. The results of these studies have been inconsistent. There are reports of increased risk, decreased risk, and no association. Differences in reference group definition, reporting bias, adjustment for confounding and effect modification are likely to have contributed to this inconsistency.

Strong association not seen. The second cause-effect standard is that there would be a strong association between having had an abortion and getting breast cancer. This standard has also not been met. Collaborative pooling and reanalysis of the data from almost all the existing studies reported a relative risk value of 1.11 for studies in which abortion information was supplied by the study participants. This value is considered to indicate a "not evident" or "very weak" increase in breast cancer risk. Some studies have sought to eliminate reporting bias and have relied on medical records or healthy reporting (as in cohort studies) of abortion and this method has produced better consistency. However, collaborative pooling and reanalysis of almost all these studies reported a relative risk of 0.93; this value is considered a "not evident" or "very weak" decrease in breast cancer risk. Of the 53 separate studies that met the criteria for the pooling reanalysis (100 or more women with breast cancer, in a country with liberal abortion laws, and having systematically sought after reproductive history), the largest reported relative risk was 1.40, which would be considered a moderate increase in breast cancer risk. A relative risk value of 1.0 represents a "not evident" association. The values from the collaborative reanalysis and other large studies, including one study of 1.5 million women in Denmark (Melbye et al., 1997), strongly suggest a lack of a relationship between abortion and breast cancer.

No dose-effect relationship. The third standard would be that a dose effect be demonstrated. This would mean that women who had more abortions would have a higher risk of breast cancer. This standard was also not met. In the pooling reanalysis, women who had two or more abortions did not have a higher risk of breast cancer than women who had one abortion. Examination of the results of individual studies that examined this issue also supports no change in risk for women who have had more abortions.

Biological plausibility exists. The fourth standard is that the effect be biologically plausible. Animal studies support the plausibility of there being a relationship between abortion and breast cancer risk (Russo et al., 1982). But this support is dependent on reference group choice. Animals whose pregnancies are terminated at about the middle of the pregnancy term (day 12 of a 21 to 23 day gestation period) have comparable numbers of tumors following carcinogen treatment and levels of proliferation to similar aged virgin rats. Levels of proliferation in the breast are related to the susceptibility of these breast cells to cancer causing chemicals. However, the levels of proliferation and tumors formed in the pregnancy-terminated rats were higher than similar aged animals that completed their pregnancies. It should also be noted that pregnancy, while ultimately decreasing breast cancer risk, is linked to a temporary increase in risk following childbirth. (Hsieh et al., 1994; Lambe et al., 1994; Liu et al., 2002) Accordingly, the time during the pregnancy when the abortion occurs may be important. Most abortions occur within eight weeks of the 40 week human gestation period, and this is a different pregnancy time period than that examined in the animals.

Time of exposure appropriate. The final standard is that the time of exposure be appropriate. This standard is met. 97% of abortions are carried out in women younger than 40 years old (CDC, 2006) and 99% of breast cancer cases occur after 40 (National Cancer Institute, SEER Cancer Statistics Review 1975-2002).

As this analysis demonstrates, only two of the five main standards for a cause-effect relationship are met. This indicates that while many studies have examined this relationship, there is little evidence to support abortion causing breast cancer. This underlies the National Cancer Institute's statement that it is "well established" that induced abortion is not associated with breast cancer risk.

The decision regarding an abortion is certainly one of the most difficult that couples or individual women ever have to make. Fear of an association of induced abortion and breast cancer should not enter into this extremely difficult decision.

References

American Cancer Society. (2005). Can Having an Abortion Cause or Contribute to Breast Cancer?

American College of Obstetrians and Gynecologists. (2003). ACOG Finds No Link Between Abortion and Breast Cancer Risk (Washington, DC).

Beral, V., Bull, D., Doll, R., Peto, R., and Reeves, G. (2004). Breast cancer and abortion: collaborative reanalysis of data from 53 epidemiological studies, including 83,000 women with breast cancer from 16 countries. Lancet 363, 1007-1016.

CDC (2006). Table 4. Reported legal abortions, by age group of women who obtained an abortion and state of occurence --selected states, United States, 2002.

Guttmacher Institute (2006). State Policies in Brief: Mandatory Counseling and Waiting Periods for Abortion (New York, NY, Guttmacher Institute), pp. 1-3.

Hsieh, C., Pavia, M., Lambe, M., Lan, S. J., Colditz, G. A., Ekbom, A., Adami, H. O., Trichopoulos, D., and Willett, W. C. (1994). Dual effect of parity on breast cancer risk. Eur J Cancer 30A, 969-973.

Lambe, M., Hsieh, C., Trichopoulos, D., Ekbom, A., Pavia, M., and Adami, H. O. (1994). Transient increase in the risk of breast cancer after giving birth. N Engl J Med 331, 5-9.

Liu, Q., Wuu, J., Lambe, M., Hsieh, S. F., Ekbom, A., and Hsieh, C. C. (2002). Transient increase in breast cancer risk after giving birth: postpartum period with the highest risk (Sweden). Cancer Causes Control 13, 299-305.

Melbye, M., Wohlfahrt, J., Olsen, J. H., Frisch, M., Westergaard, T., Helweg-Larsen, K., and Andersen, P. K. (1997). Induced abortion and the risk of breast cancer. N Engl J Med 336, 81-85.

National Cancer Institute. (2003). Summary Report: Early reproductive events and breast cancer workshop (Bethesda, MD, National Cancer Institution).

National Cancer Institute. (SEER Cancer Statistics Review 1975-2002). Table IV-4. Breast Cancer (Invasive) U.S. Death Rates, Age-Adjusted and Age-Specific Rates, By Race and Sex.

Royal College of Obstetrians and Gynecologists. (2004). RCOG Statement on Abortion and Breast Cancer.

Russo, J., Tay, L. K., and Russo, I. H. (1982). Differentiation of the mammary gland and susceptibility to carcinogenesis. Breast Cancer Res Treat 2, 5-73.

World Health Organization. (2000). Induced abortion does not increase breast cancer risk (Geneva, Switzerland).

Back to the top

Last Update