In attempting to explain why these efforts at consensus generated such antagonism, various commentators have convincingly argued that these debates were not really about the scientific value of mammography. Indeed, it has been claimed that there is broad agreement on what the data show [4]. Instead, as Fletcher [5], Ernster [6], and others [7, 8] have asserted, public acrimony reflected the entrance of political, economic, legal, and interest group concerns into the screening process [9]. A related issue has received less attention: how the language of advocacy may itself polarize scientific discussion, leaving physicians and patients without adequate guideposts for applying early detection of breast cancer to clinical practice.

I argue that the rhetoric of a decades-old "war" against breast cancer framed the recent arguments about breast cancer screening. Although efforts to control other cancers and other diseases have also used military metaphors, breast cancer provides one of the most vivid examples of how metaphoric language enters scientific debate and, in turn, may influence such debate.

This paper will examine two historical controversies: 1) debates over whether early detection of cancers improved survival and 2) debates over biologically indeterminate "precancers" that early detection often revealed. By revisiting the role of military metaphors, history can help explain why these debates have been so divisive and why the value of early detection has, at times, been oversold. This analysis has direct implications for genetic testing, where the next "battle" is already being fought.

The War on Breast Cancer

More than 20 years ago, Susan Sontag pointed out that disease metaphors were not simply words but rather that they often acquired a "striking literalness and authority" [10]. In the case of cancer, the use of war metaphors implied that the disease was an actual enemy to be vanquished on a medical battlefield. Such military language has been particularly prominent in breast cancer.

In 1936, activists formed the Women's Field Army, whose "war cry" was for "trench warfare with a vengeance against a ruthless killer" [11, 12]. Although the Women's Field Army focused on all cancers in women, breast cancer generated increasing interest and controversy in the years after World War II. Not only did breast cancer kill more women than other cancers, but its location in an external organ made efforts at "tracking down the enemy" seem promising and urgent [13]. Moreover, given the association of women's breasts with intimacy and sexuality, detection and removal of breast cancer caused particular fear and anxiety. As one writer stated, there was "a pathological national anxiety bordering on hysteria" about breast cancer [13]. In such a setting, military imagery gave force and direction to screening and treatment programs. Victory over breast cancer, physicians remarked, required "a carefully planned military campaign" [14] and an "increase [in] the caliber of our weapons" [15].

Whereas breast lumps had previously been discovered by accident, efforts after 1945 urged both women and physicians to find hidden masses through breast examination. Hands used in this manner became "weapons" [13]. This new strategy stemmed from the longstanding theory that breast cancer began as a tiny focus that grew locally in a predictable and gradual manner before spreading. Data showed that stage I cancers (those confined to the breast) carried a better prognosis than those involving axillary nodes or other organs [14]. These statistics encouraged the American Cancer Society to promote early detection of these small, presumably localized lumps [16].

If the value of early detection was a given, so was the surgery that followed: radical mastectomy. Popularized by Halsted in the late 1800s, the operation reflected the centrifugal model of cancer spread and the belief that curative surgery required removal of all cancer cells [17, 18]. Radical mastectomy entailed resection of the breast, surrounding tissues, and both pectoral muscles and complete axillary node dissection. This extensive and disfiguring operation was recommended even for small, apparently localized tumors [19].

Debates over Early Detection

Beginning in the 1950s, controversy arose over the early detection of small breast cancers. Directly challenging the existing paradigm, some commentators claimed that early detection and aggressive treatment had little effect on the natural history of breast cancer. Instead, they argued, biological factors, such as tumor virulence and immune response, determined the fate of patients.

MacDonald first proposed this theory of "biological predeterminism" in 1951, on the basis of research that revealed no consistent association between the extent of breast cancer and the time that elapsed before a woman showed a detected lump to a physician. Using a somewhat biased sample, Mac-Donald also found that the size of the primary breast lesion was not reliably related to distant disease: Fifty-six percent of tumors 1 cm or smaller had positive axillary nodes, and 23% of tumors larger than 5 cm had negative nodes [20].

A group of biometricians extended MacDonald's conclusions, arguing that the 70% 5-year survival rates (the standard at the time) credited to early detection and radical surgery were somewhat inflated [21-24]. That is, physicians were inappropriately counting as "cures" patients whose cancers, if left undetected, would not have killed them. As McKinnon wrote, "Curing non-lethal lesions does not reduce mortality" [23]. This work presaged such biostatistical concepts as lead-time and length biases [25]. National data seemed to confirm McKinnon's argument: Despite 50 years of radical surgery, the breast cancer mortality rate-roughly 25 per 100 000 persons-remained essentially unchanged [21-23].

This stationary death rate led biological predeterminists to term the association of early treatment and curability a "shibboleth" [20]. Black and Speer [22] claimed that without a reliable connection among delay in diagnosis, extent of disease, and curability, it was wrong to equate early lesions with small size and lack of metastasis. Rather, "biological type-the propensity of spread and development of remote metastases-plays the predominant role in determining the outcome" [26]. Thus, stage I tumors did better not because of early detection and removal but because most of them were "non-progressive."

Furthermore, small lesions that metastasized early challenged the standard view of how breast cancer spread. "The majority of so-called operable cases of breast cancer," Black and Speer argued, "have already undergone occult dissemination at the time of surgery" [22]. In turn, this conclusion called into question the need to remove all cancerous tissue. Eventually, controlled trials demonstrated that radical mastectomy was unnecessary [17].

Despite their seemingly gloomy message, the biological predeterminists were hardly nihilists. Even MacDonald remarked that the prognosis of 25% of breast cancers could be improved through early detection and radical mastectomy [27]. Nevertheless, many predeterminists were prone to hyperbole. For example, Lees, after summarizing the existing paradigm of breast cancer, wrote "Now this is all non-sense and contradicted by practical experience" [26]. Crile accused anticancer agencies of creating "a new disease, cancer phobia," that "causes more suffering than cancer itself" [28].

Opponents of biological predeterminism responded with perhaps greater intolerance, seeking to "demolish" even reasonable claims of the predeterminists [29]. A 1952 presentation by Lees provoked Taylor to suggest that his colleagues might want to "rend [Lees] limb from limb" [26]. Referring to McKinnon, Taylor stated that "the devil can quote scripture to his purpose" [26].

Behind this banter lay intense frustration and anger that discouraged rational analysis [30, 31]. Even as MacDonald's opponents contested his data, they were especially perturbed that someone had challenged standard cancer-control philosophy. As several physicians wrote in response to Crile's call for less radical surgery, "Dr. Crile offers a dangerous, fatalistic philosophy of cancer" [32].

The "war on cancer" metaphor had important benefits, both in publicizing breast cancer and in generating substantial research funding [33, 34]. However, it discouraged physicians and patients from acknowledging the ambiguous results that early detection often produced; limited warfare held little appeal. A similarly aggressive attitude would characterize debates about lesions that were not cancers at all.

Debates over "Precancers"

Physicians had long identified hyperplastic breast lesions that had not invaded the basement membrane [35]. By the 1930s, pathologists, believing that such lesions were "precancers," began to term them ductal or lobular "carcinoma in situ" [36]. In this section, I review debates about the lobular variant. Lesions revealing lobular carcinoma in situ were of limited concern before 1950. Because they were nonpalpable, they were discovered only incidentally, usually during histologic examination of benign nodules. With the growing emphasis on early detection and biopsy of suspicious lumps, diagnosis of lobular carcinoma in situ increased substantially by the 1970s [37]. Since the widespread diffusion of mammography after 1975, detection of in situ carcinomas, particularly the ductal type, has become even more commonplace [38, 39].

Lobular carcinoma in situ raised two questions: 1) Were such lesions inevitably precancerous? 2) Did detection mandate mastectomy? Again, debate ensued. A few surgeons favored a conservative approach, calling such lesions "lobular neoplasia" and advising observation [40]. However, most physicians, using familiar military metaphors, were aggressive. As one physician asked, "Does the treatment of this deceptive disease demand the broadsword or will the rapier suffice?" [41]. Ultimately viewing carcinoma in situ as a "powder keg" [42], surgeons generally recommended mastectomy [43, 44]. This aggressive use of surgery demonstrates how physicians, understandably concerned about leaving behind potentially malignant lesions, essentially equated breast cancer predisposition with the actual disease.

The issue of predisposition had arisen before. It had long been known that women with strong family histories were more likely to develop breast cancer. Moreover, women with past breast cancer were 4 to 10 times more likely than other women to develop cancer in the other breast, although some of these tumors were probably metastases [45, 46]. Despite such figures, prophylactic removal of the second breast was rarely considered.

Strategies changed after 1965, as physicians published data on a few patients in whom lobular carcinoma in situ was treated only by local excision. Reported rates of future invasive disease in the affected breast were as high as 35% after 20 years [47]. Although some critics urged caution, all agreed that lobular carcinoma in situ was a strong marker for future invasive cancer.

How did an excised in situ lesion predispose to future cancer? The presumed answer was multicentricity [48]: As many as 70% of breasts removed for a single lesion had multiple foci of invasive or noninvasive lobular disease [49]. Future cancers in unremoved breasts, therefore, probably arose from these foci. Given such findings, most physicians readily agreed that "lobular carcinoma in situ is a preinvasive form of breast cancer that should be treated by ... mastectomy" [50]. This new ability to "cure" precancerous lesions represented an enormous triumph for physicians ardently devoted to preventing metastatic breast cancer.

Success in treating unilateral lobular carcinoma in situ in one breast led to scrutiny of the other breast [51]. Studies reporting high rates of future ipsilateral breast cancer had also found rates of future contralateral cancer as high as 25% [47]. Once physicians began to view lobular carcinoma in situ as a bilateral disease, it became entirely "logical" [52] to pursue its early detection in the contralateral breast. Mammography was most often used, but some surgeons recommended a more aggressive approach: random biopsy of the second breast [49, 53-56]. Such biopsies were of high yield, generating rates of contralateral carcinoma in situ or cancer up to 59% [49, 53, 55]. Because mastectomy was usually undertaken for the original in situ lesion, surgeons-with the consent of their patients [51]-performed a second mastectomy if biopsy results were positive. Although few surgeons actually performed random breast biopsies and double mastectomies for bilateral lobular carcinoma in situ, the strategy received favorable coverage [43, 49, 54].

Some surgeons pushed early detection and treatment even further. Noting that lobular disease was often bilateral and that random biopsies probably missed existing lesions, they proposed routine prophylactic removal of the second breast in women with lobular carcinoma in situ [54, 57-59]. Prophylactic mastectomy, they argued, was "almost total breast biopsy" and thus of "great merit in the discovery of occult carcinoma" [58]. It might be argued that the war on breast cancer reached its pinnacle when a marker of possible future cancer in one breast became a rationale for bilateral prophylactic mastectomy in women without other risk factors.

By the mid-1970s, the use of mastectomy for lobular carcinoma in situ was being challenged [52, 60]. Noting that most women treated only with lumpectomy never developed cancer and that those that did frequently had ductal carcinoma, commentators argued that "random biopsy specimens ... detect morphological entities that will not be manifested as clinical disease in a normal lifetime" [61]. Gradually, most physicians began to recommend only close observation, although some women today choose prophylactic mastectomy [62, 63].

Conclusion: Lessons for Genetic Testing

Building on arguments that the aggressive use of radical surgery, radiotherapy, chemotherapy, bone marrow transplantation, and even chemoprophylaxis has often done more harm than good [64-66], critics have contended that the war on breast cancer has achieved little [67, 68]. Others disagree [69, 70], citing improved disease-free survival and the recent decrease in overall breast cancer mortality [71]. This paper does not attempt to resolve this dispute. Instead, it seeks to demonstrate how military metaphors, by polarizing scientific discussion, have interfered with efforts to provide clinicians and patients with useful recommendations on early breast cancer detection.

Excessive hyperbole characterized the initial response to the January 1997 NIH decision to oppose routine mammography for women 40 to 49 years of age [1-8]. (The National Cancer Institute later recommended regular mammography for these women.) Even though the panel had carefully reviewed extensive data, commentators termed its findings "fraudulent" and "tantamount to a death sentence" for women [2]. Former NIH director Bernadine Healy even upbraided the panel for daring to challenge the ethos of early detection [2].

It is not surprising that such language accompanied debates over breast cancer. Although similar rhetoric may characterize discussion of other cancers, issues surrounding the detection and treatment of breast cancer have consistently raised high emotion among advocacy groups. Arguments about the benefits and dangers of mammography in the 1970s made the front pages, pitting scientist against scientist [72]. In the same era, surgeons and newly empowered women engaged in highly charged debates over radical mastectomy and routine adjuvant chemotherapy [66, 73]. State legislatures eventually became involved, passing laws to ensure informed consent for patients with breast cancer [74]. Legal, political, economic, and gender considerations continue to influence breast cancer screening and treatment. For example, on learning of the results of the January 1997 consensus conference, the U.S. Senate voted 98 to 0 in favor of screening mammography for women 40 to 49 years of age [3].

Given the concerted efforts of advocacy groups to influence breast cancer policies, the frequent use of war metaphors is not surprising. What began as metaphorical language has become "received wisdom" [75], and it is now almost impossible to discuss breast cancer without using military terminology. For example, a recent article described a breast cancer activist as a "modern-day warrior doing battle for women" [76]. United States Senator Olympia Snowe recently urged Americans "to wage war against a brutal and merciless enemy: breast cancer" [77]. Books, Web sites, and the media routinely characterize breast cancer control efforts as a "fight" or "battle" [3, 78-84].

If the war on breast cancer has achieved its successes, it has also contributed to a highly contentious atmosphere in which the actual value of early detection and aggressive treatment has at times been oversold. This lesson is especially important as breast cancer moves to a "genetic battlefield" [85]. Tests identifying BRCA1 and BRCA2 genetic mutations, which are already being termed "time bombs" [86], represent a powerful new method for obtaining early information about potential breast cancer [87-92]. Advertisements for these tests feature women saying, "Doctor, I need to know." However, lacking knowledge about the actual value of these tests, history warns us not to conflate the ability to find these markers with the need to find and act on them.

History also informs us about the consequences of identifying healthy women with an increased predisposition for breast cancer. Women who test positive for BRCA1 or BRCA2 have been termed "unpatients" [93], not diseased but at risk for disease. As in the case of lobular carcinoma in situ, such a distinction may easily be blurred: Both physicians and patients may equate cancer predisposition with actual cancer. This message is particularly important as women with genetic mutations consider the often reasonable option of prophylactic mastectomy [94]. In the past, an "all-or-nothing" wartime mentality favoring definitive early intervention fostered overuse of this procedure.

The encouraging recent decline in breast cancer mortality will probably accelerate calls to fight the disease, but we should recall the consequences of all-out war.

Barron H. Lerner, MD, PhD

Columbia University; New York, NY 10032