Debunking the Abortion-Breast Cancer Hypothesis – the wrap-up show.

Since the publication of the Collaborative Group on Hormonal Factors in Breast Cancer’s meta-analysis of Aborton-Breast Cancer (ABC) studies in 2004 – Beral et al. (2004) [1] – two further large prospective studies have been published, neither of which found that induced abortion was associated with an increase risk of developing breast cancer.

The Harvard Nurse’s Health Study – Michels et al. (2007) – included 105,716 women aged 29-46 years old at the start of the study in 1993 generating 973,437 person-years of follow-up between 1993 and 2003. Information on induced and spontaneous abortions and breast cancer diagnoses were collected by self-report questionnaire in 1993 and updated every two years during the study period. Self-reported breast cancer was histologically confirmed for 99% of women whose medical records could be obtained and because the accuracy was so high, self-reported cancers were included where these could not be confirmed by medical records,  The study reporting the following results:

A total of 16 118 participants (15%) reported a history of induced abortion, and 21 753 (21%) reported a history of spontaneous abortions. The hazard ratio for breast cancer among women who had 1 or more induced abortions was 1.01 (95% confidence interval, 0.88-1.17) after adjustment for established breast cancer risk factors; among women with 1 or more spontaneous abortions, the covariate-adjusted hazard ratio was 0.89 (95% confidence interval, 0.78-1.01). The relation between induced abortion and the incidence of breast cancer did not differ materially by number of abortions (P for trend = .98), age at abortion (P for trend = .68), parity (P for interaction = .54), or timing of abortion with respect to a full-term pregnancy (P for interaction = .10).

The California Teachers Study – DeLellis Henderson et al. (2008) – used a similar design with induced and spontaneous abortion data and historical cancer data being obtained prospectively via self-report questionnaires while data for newly-diagnosed cancers was taken from the California Cancer Registry. Women with a prior history of breast cancer of whose history was unknown were excluded from the stud, as were women who were 80 years of age or older at the start of the study, women who were currently pregnant with their first pregnancy  and women who had incomplete information on critical breast cancer risk factors, such as unknown age at menarche, unknown if ever pregnant, unknown number of pregnancies, etc. 109,853 women were, therefore, included in the study cohort, of which 61,00o were post-menopausal, with follow-up lasting nine years. The study reported a relative hazard ratio for induced abortion of 0.95 (CI 95% 0.76–1.18) for nulliparous women and 1.05 (CI 95% 0.92–1.20) for parous women. Sub-group analyses for number of induced abortions, age at first induced abortion and whether the abortion occurred prior to or following legalisation (1973) all produced non-significant results and, therefore, no association between induced abortion and breast cancer.

Brind and Lanfranchi have, inevitably, criticised both studies in an effort to discredit their findings, in fact both were subjected to much the same criticisms

– allegedly inadequate follow-up time, which is controlled for in both studies by age stratification and therefore not the problem that Brind would like people to think it is, and

– exclusion of women diagnosed with ductal carcinoma in situ (DCIS), which Lanfranchi attacks in the following terms:

After 2005, there were only two papers published — in 2007 and 2008 — that concerned only abortion and breast cancer, the Harvard Nurses Study and the California Teachers Study. Both studies concluded that there was no increased risk of breast cancer with abortion. Disturbingly, both studies did not include the development of ductal in-situ breast cancers which account for over 60,000 cases of breast cancer a year. In situ breast cancer is treated with surgery, radiation and drugs. It has increased in incidence in premenopausal women 400% over the last 40 years in the US and 600% worldwide. Furthermore, most of these cancers develop into invasive breast cancers, which are the only cancers these two studies considered. If a cancer is serious enough that a woman might need to undergo mastectomy in order to be cured, as is so often the case with in situ cancers, it is strange to have kept the count of such cancers out of the study. By disregarding what the public and all women consider to be serious threats to their lives (in-situ breast cancers) the results are not reflective of a woman’s true cancer risk, nor of women’s desire to know the truth, nor of the true rate of breast cancer.

Ductal carcinoma in situ is the most common type of non-invasive breast cancer (or perhaps pre-cancer) in which cancer cells develop in the milk ducts but are found to have remained in place, i.e. they haven’t invaded other breast tissue at the point of discovery. Contrary to Lanfranchi’s simplistic and polemical, if not slight hysterical, remarks, DCIS is a much more complex matter than her diatribe suggests.

DCIS is not a single disease but rather a complex family of diseases which can range from non-life threatening low grade lesions to high grade lesions that may contain the foci of invasive breast cancer and is also almost always asymptomatic, i.e. it doesn’t produced symptoms or a lump which can be felt externally, which is why it is always always diagnosed through mammography screening.

The problem with including DCIS in ABC studies is perhaps best explained in an editorial by Welch et al [4], which was published in 2008:

Wouldn’t it be easier if we lived in a binary world? Everything would be either black or white, yes or no, 1 or 0. And biopsy results would be either normal or cancer.

Unfortunately, in the world of cancer, our efforts to detect the disease early have made this a fantasy. We are increasingly faced with the reality of a big gray zone — a broad spectrum of pathologic findings between normal tissue and invasive cancer. And our nosology reflects the associated ambiguity — dysplasia, intraepithelial neoplasia, hyperplasia with atypia, and even abnormalities of “unknown significance.” The unifying theme for these findings is that they may progress to invasive cancer. Or they may not.

Despite the presence of the word “carcinoma,” ductal carcinoma in situ (DCIS) is the poster child for this problem (a senior pathologist involved in developing classification systems confided to one of us that he regretted the use of the term carcinoma in DCIS). No one believes that DCIS always progresses to invasive cancer, and no one believes it never does. Although no one is sure what the probability of progression is, studies of DCIS that were missed at biopsy and the autopsy reservoir suggest that the lifetime risk of progression must be considerably less than 50%.

There is an added complexity: DCIS is associated with not just one risk but two. In addition to the specific risk that the lesion might progress to invasive cancer, DCIS confers a general risk. It is a marker for an increased chance of developing invasive cancer elsewhere in the ipsi- or contralateral breast. Again, although no one is sure what this probability is, a recent prospective study of a cohort of patients with DCIS who were treated largely by excision alone suggested that the 5-year risk of subsequent invasive breast cancer elsewhere is less than 10%.

In short, there is a sea of uncertainty surrounding DCIS. Some lesions will progress to cancer, others will not. Some women with DCIS will develop cancer elsewhere in their breasts, whereas others will not. And we’re not sure what the chances are.

Lanfranchi claims that most diagnosed cases of DCIS into invasive breast cancers when, in reality, studies of cases missed at biopsy and tissue obtained at autopsy suggest that the lifetime risk of progression for DCIS ‘must be considerably less than 50%’ while the 5-year risk of breast cancer elsewhere after excision (removal) of DCIS lesions is less than 10%.

In reality, no one currently knows what the odds of DCIS developing into invasive breast cancer are nor, on finding DCIS lesions via mammography, can oncologists predict which one’s will result in invasive breast cancer and which one’s won’t – and so, inevitably, doctors – and women, of course – will almost always take the safe option of either a lumpectomy or, in some cases, a mastectomy even though there is no way of knowing whether such disfiguring treatment was actually necessary.

Lanfranchi also notes the increase in incidence of DCIS over the last 40 years, which all too conveniently invite a comparison with the period during which abortion has been legal in all US States – 39 years and counting –  following the Supreme Court ruling in Roe vs Wade, but makes no mention of the actual reason for this increase, which is the introduction and widespread use of mammography screening. Although there are, to date, no US studies, which look at the impat of screening on DCIS incidence rates, a Dutch study by van Steenbergen et al. (2008) [5] found that most of the increase in DCIS incidence in the Southern Netherlands in women over 50 is accounted for by the introduction of mammography screening while for women under 50 the rising incidence of DCIS ‘may be attributed largely to the more
vigorous search for families with heritable breast cancers in the Netherlands, following the identification of BRCA1 and BRCA2 mutations since 1995″, with increased public awareness and improved detection methods also making a contribution.

In short, the more effort you put into looking for cancers, the more cancers you’ll find, pushing up incidence rates in the process.

Throw in problems with overdiagnosis:

Long-term follow-up of the Malmö randomized tria suggests that a quarter of mammographically detected breast cancers represent overdiagnosis. Although it is impossible to determine which individuals are overdiagnosed (unless they are not treated and ultimately die without ever developing symptoms from their cancer), it is possible to identify subsets of patients who are at high risk of overdiagnosis. In breast cancer, this subset is patients with DCIS. Because the “best guess” is that most DCIS won’t progress to invasive cancer, the risk of overdiagnosis would be expected to be greater than 50%.

And the reason that women diagnosed with DCIS were excluded from these two studies is glaringly obvious. There is absolutely no way of reliably estimating how many of the women with DCIS would go on to develop an invasive breast cancer or of accurately linking such estimates to the abortion data. Nevertlhess, Michel et al. make the observation that:

We censored cases of carcinoma in situ (n = 399) from the primary analyses, but results including in situ cases were comparable to those for invasive cases only.

A secondary analysis which, in keeping with Brind and Lanfranchi’s treatment of Beral et al. they fail to report.

There have also been a number of small retrospective studies published since 2004, all of which are reported uncritically by Brind and Lanfranchi because they report a positive association between abortion and breast cancer, even if one of these studies in misidentified by Lanfranchi as originating from Albania when it is, in fact, an Armenian study.

Since publishing his own 1996 meta-analysis Brind has confined his in-print activities largely to critiquing, or perhaps more accurately, attacking studies which report findings that fail to support the ABC hypothesis via letters to the journals that published those studies. Up until 2004, these critiques were invariably presented as having been co-authored with Vernon M Chinchilli, the statistician from Penn State University who collaborated with Brind on his 1996 paper however that association appears to have ceased in 2004, leaving Brind to fly solo, and may well have contributed to a noticeable decline in Brind’s ability to get his critical comments accepted for publication by credible medical  journals.

Brind has published critiques of Beral et al and of the Harvard Nurses and California Teachers studies but only in JPandS (The Journal of American Physicians and Surgeons) which is not listed in Medline/Pubmed or Web of Sciencenor indexed by the US National Library of Medicine – JPandS is the Conservapedia of medical journals. The Breast Cancer Prevention Insitute’s website also includes a PDF document contaiing Brind’s critique of Beral et al. which is referenced on the site as “Abortion and Breast Cancer: Re: “collaborative reanalysis of data” Lancet. 3/25/2004″. To a lay observer that ‘reference’ could be taken as an indication that the lette was published in The Lancet – it wasn’t, although its not clear whether this is because it was never sent to The Lancet or whether they simply – and sensibly – decline to publish it.

And that’s pretty much it for the evidence base as it relates to the ABC hypothesis – the best available evidence we have does not show a link between abortion and breast cancer – which leaves me only to offer a few thoughts on the question of why its possible that the ABC hypothesis may well being to resurface in the coming months.

In one sense, the ABC hypothesis has never really gone away. Although it not accepted by of the major cancer research organisations, professional OB/GYN bodies or the World Health Organisation, it continues to be promoted by Brind and Lanfranchi, through their ‘Breast Cancer Prevention Institute’ and by a wide range of other anti-abortion lobby organisations while at least six US states, including Mississippi and Texas have ‘informed consent’ laws which compel doctors to give women seeking an abortion inaccurate and misleading information about abortion  and breast cancer.

In the UK, a mystery shopper-based investigation of 10 ‘crisis pregnancy centres’ operated by Christian/Anti-abortion organisation found that literature suggesting that abortion was linked to increased risk of breast cancer was being given to women using its services:

At Skylight Counselling, a Care Confidential affiliate in Coventry, the counsellor was said to have listed physical and psychological effects she linked to the “post-traumatic stress” of abortion. Literature suggesting risks such as “sub-fertility or infertility” and “increased risk of breast cancer” was given out.

With a consultation on possible changes to the framework under which pre-abortion counselling is offer to women seeking an abortion in England and Wales in the offing there are concerns that any changes to the present system could open the door not only to much greater involvement from Christian-run CPCs and result in women being given inaccurate information about risk associate with abortion should any revised framework not include a specific requirement that ties information provision explicitly to the evidence-based guidelines issued by the Royal College of Obstetricians and Gynaecologists.

If the ABC hypothesis does begin to resurface in the coming months – and it may not – the most likely reason for a renewed interest in it amongst anti-abortion groups in the UK may stem from last year’s systematic review of the evidence relating to abortion and mental health, which firmly rejected the claim that induced abortion places women an a greater risk of developing mental health problems. In recent year, the anti-abortion lobby in the UK has invested heavily promoting claims of a link between abortion and mental health in the hope that it would provide a means of trying influence women against having abortion and justify demands for supposedly ‘independent’ pre-abortion counselling, much of which would be provided via their own CPCs, only to have the rug pulled from under them by the RCPsych review and, more recently, the discrediting of a published study by Priscilla Coleman in which she was found to have misreported the dataset on which her study was based, after being challenged publicly by other researchers, and to have carried out an invalid analysis based on lifetime risk data.

With the Abortion-Mental Health hypothesis have proved to be a busted flush, some parts of the anti-abortion lobby may well return to the ABC hypothesis in the hope that this may provide them with the clinical argument they’re seeking to promote their agenda, which may co some way toward explaining why a commentary on Lanfranchi’s ‘paper’ turned up in a recent issue of the Catholic Herald.

At that, as they say, is that – although I may try and put together a short and more easily digestible summary of the evidence in the near future, now repeat after me…

…the totality of the worldwide epidemiological evidence indicates that pregnancies ending as either spontaneous or induced abortions do not have adverse effects on women’s subsequent risk of developing breast cancer.

To read the other articles in this series use the following links:

Debunking the Abortion-Breast Cancer Hypothesis – pt1.
Debunking the Abortion-Breast Cancer Hypothesis – pt2.
Debunking the Abortion-Breast Cancer Hypothesis – pt3.
Debunking the Abortion-Breast Cancer Hypothesis – pt4.

References.
1. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and abortion: Collaborative reanalysis of data from 53
epidemiological studies, including 83,000 women with breast cancer from 16 countries. 2004;363:1007-1016.

2. Michels K, Xue Fei, Colditz G., Willett W. Induced and Spontaneous Abortion and Incidence of Breast Cancer Among Young Women. Arch Int Med 2007 167:814-820.

3. DeLellis Henderson K, Sullivan-Halley J, Reynolds P, Horn-Ross P, Clarke C, et al. Incomplete pregnancy is not associated with breast cancer risk: the California Teachers Study. Contraception 2008;77:391-396.

4. Welch GH, Woloshin S, Schwartz LM. The sea of uncertainty surrounding ductal carcinoma in situ—the price of screening mammography. J Natl Cancer Inst 2008;100:228-9.

5. van Steenbergen LN, Voogd AC, Roukema JA, Louwman WJ, Duijm LE, Coebergh JW, Poll-Franse LV: Screening caused rising incidence rates of ductal carcinoma in situ of the breast. Breast Cancer Res Treat 2009, 115:181-3