2023年5月9日，美国预防服务工作组USPSTF将乳腺钼靶筛查的起始年龄推荐意见从50岁提前至40岁[1]。此前，USPSTF认为对40至50岁女性进行筛查为个人选择。由于USPSTF推荐意见的影响力巨大，因此针对40多岁女性的乳腺钼靶检查可能会成为医疗保健绩效衡量标准；如果这样，将有效地成为初级保健从业者必须遵守的公共卫生当务之急。这样的变化将影响超过2000万美国女性，并引发一些重要问题。 2023年9月16日，国际四大医学期刊之一、美国麻省医学会《新英格兰医学杂志》在线发表达特茅斯研究所和达特茅斯癌症中心、丽莎施瓦茨医学真理基金会、考科蓝合作组织丹麦分中心和欧登塞循证医学中心、南丹麦大学临床研究系、杜克大学外科学系、哈佛大学布莱根医院和波士顿妇女医院的专家意见，对USPSTF新版乳腺钼靶筛查推荐意见提出不同看法。 首先，是否有新证据表明乳腺癌死亡率正在增加？相反，美国的乳腺癌死亡率稳步下降——这是现代医学的重大成功故事之一。根据全国生命统计系统的数据，50岁以下女性的乳腺癌死亡率下降最为明显，过去30年来她们的乳腺癌死亡率降低了一半。其他高收入国家也出现了类似的情况，包括对40多岁女性极少进行筛查的国家（丹麦和英国），以及对全部年龄组进行筛查的国家（瑞士），这表明筛查率下降很大程度上是由于治疗方法的进步，而非筛查，参见下图。4个筛查做法不同国家的乳腺癌死亡率变化趋势 其次，是否有新的证据表明乳腺钼靶检查的获益正在增加？自从USPSTF提出之前的推荐意见以来，还没有针对40多岁女性进行乳腺钼靶检查的最新随机对照试验。针对该年龄组的8项随机对照试验，包括最近的英国Age研究，都表明未见显著效果[2]。该发现既反映了40多岁女性乳腺癌相关死亡的罕见性，也反映了筛查降低死亡率的程度低于预期的事实——也许是因为该年龄组发生了恶性程度更高的病变。快速生长的癌症更容易被筛查漏检，通常出现于两次筛查之间。 新的推荐意见并非根据新的试验数据，而是根据统计学模型，该模型估计如果降低起始年龄可能发生什么。这些模型假设乳腺钼靶检查可将乳腺癌死亡率降低大约25%，并得出结论：对40～74岁（而非50～74岁）女性进行筛查，终身乳腺癌死亡率将降低1‰～2‰。 USPSTF日益依赖复杂统计学模型是有问题的。估计的效果可能对建模假设极其敏感，这通常反映了当时的传统智慧。在女性健康倡议研究之前进行的一项著名模型预测，几乎全部绝经后女性的预期寿命都将通过激素替代疗法延长，结果被女性健康倡议研究否定。模型可能具有明显的定量精度吸引力，但是其可靠性取决于输入数据和假设。正如其他人所说，政策制定者仅在了解模型背后的参数和假设时才应该采用模型[3]。 在该情况下，尤其有问题的是，统计学模型推算出通过乳腺钼靶筛查可将乳腺癌死亡率相对风险降低25%超过了随机对照试验荟萃分析观察到的结果：全部8项试验合计相对风险降低16%（95%置信区间：-27%～-4%），3项低偏倚风险试验的相对风险降低13%（95%置信区间：-27%～+3%）[2]。 那么，获益和危害的平衡是否支持新的公共卫生要求？相对风险降低可能会产生误导，因为其中不包含关于绝对风险的信息，而对于该年龄组而言，绝对风险已经很低并且正在稳步下降。为了从绝对意义上阐明更新推荐意见的潜在影响，下表总结了获益和危害。美国40多岁女性每2年进行1次乳腺钼靶检查的利弊 
对于40多岁的美国女性而言，无论是否进行筛查，未来10年内全因死亡风险约为3%。乳腺钼靶检查的模拟获益是将女性10年乳腺癌死亡风险从大约0.3%降低至大约0.2%，相差0.1个百分点（每1000位接受10年筛查的女性有1人死于乳腺癌）。换而言之，通过筛查，未来10年不死于乳腺癌的可能性从99.7%增加到99.8%。 该影响微乎其微，尤其考虑到潜在的危害以及似乎过于乐观的获益假设。到目前为止，最常见的结局都是误报：USPSTF模型估计，40～49岁女性有36%在10年内每两年一次筛查过程中至少进行过一次筛查。所有人都需要更多检测以证明她们未患癌症；有些将接受多次检测并面临巨额自付费用。有些人将经历恐惧：大约三分之一的女性将这种经历描述为“非常可怕”或“我一生中最可怕的时刻”[4]。 大约6.6%的接受筛查女性被误报需要进行活检。此外，USPSTF模型估计，0.2%的癌症患者可能由于本来不会造成危害的癌症而被过度诊断和治疗。这些危害在实践中可能更多，因为来自乳腺癌监测联盟的模型输入可能仅反映高绩效实践的发生率。如果每年而非像大多数美国女性目前实际每两年进行一次筛查，那么危害将更多。 上表反映了40多岁女性面临的重要抉择：少数女性的获益是否大于影响更多女性的危害？答案是女性应该能够为自己做出价值判断，而非医生强加给她们的公共卫生义务，以达到“质量”指标。由于过去30年来治疗方法的进步，死亡率稳步下降，随着时间的推移，对筛查获益的女性可能越来越少，而更多的筛查可能增加危害。 尤其USPSTF还认为，新版推荐意见是缩小黑人与白人女性乳腺癌死亡率差异的重要第一步。根据国家癌症研究所监测流行病学最终结果数据库，虽然自1990年以来，两组40多岁女性的死亡率都下降大约一半，但是该差异令人不安且持续存在：根据全国生命统计系统的数据，黑人女性死于乳腺癌的比例（10万分之23）仍然远高于白人女性（10万分之13）。但是很难想象向这两个群体推荐相同的干预措施将如何缩小差距，尤其考虑到40多岁的黑人与白人女性相比，筛查率已经同样高（根据全国卫生统计中心的数据，大约60%）。 更多的筛查无法解决癌症生物学的根本差异：根据国家癌症研究所的数据，黑人女性三阴性乳腺癌发病率是白人女性的两倍，该亚型恶性程度最高，治疗效果最差，并且最有可能被筛查漏检[5]。 早期筛查也无法解决贫困女性所面临的问题，这些女性往往黑人比例过高，例如可获得的医疗服务质量较低、对扫描结果异常的随访延迟、治疗延迟以及术后辅助治疗使用较少。事实上，降低筛查年龄实际上可能加剧造成差异的问题——因为资源将被转移到扩大筛查范围。我们需要做更多真正有效的事情：确保贫困的乳腺癌女性更容易获得高质量治疗。 如果有证据表明乳腺癌结局正在恶化，或者有新证据表明筛查年轻女性获益明确，那么乳腺钼靶检查推荐意见的改变就会得到支持。事实上，这两个前提条件都不存在。 我们希望政策制定者重新考虑降低乳腺钼靶检查起始年龄的决策。由于其获益有限以及对健康女性如此常见和重要的危害，USPSTF的模型不足以支持新的公共卫生要求。最好允许女性根据自己对数据和价值观的评定自己决策，并重新分配资源以确保全部乳腺癌女性都能得到最好、最公平的治疗。Preventive Services Task Force. Draft recommendation statement — breast cancer: screening. May 9, 2023. www.uspreventiveservicestaskforce.org/uspstf/draft-recommendation/breast-cancer-screening-adults Gotzsche PC, Jorgensen KJ. Screening for breast cancer with mammography. Cochrane Database Syst Rev. 2013;2013(6):CD001877-CD001877. Kramer BS, Elmore JG. Projecting the benefits and harms of mammography using statistical models: proof or proofiness? J Natl Cancer Inst. 2015;107(7):djv145-djv145. Schwartz LM, Woloshin S, Fowler FJ Jr, Welch HG. Enthusiasm for cancer screening in the United States. JAMA. 2004;291:71-78. Hayse B, Hooley RJ, Killelea BK, Horowitz NR, Chagpar AB, Lannin DR. Breast cancer biology varies by method of detection and may contribute to overdiagnosis. Surgery. 2016;160:454-462.
相关链接 N Engl J Med. 2023 Sep 16. IF: 158.5 The New USPSTF Mammography Recommendations — A Dissenting View. Steven Woloshin, Karsten Juhl Jorgensen, Shelley Hwang, H. Gilbert Welch. Dartmouth Institute and Dartmouth Cancer Center, Lebanon, NH; Lisa Schwartz Foundation for Truth in Medicine, Norwich, VT; University of Southern Denmark, Odense; Duke University, Durham, NC; Brigham and Women's Hospital, Boston. Recently, the U.S. Preventive Services Task Force (USPSTF) changed its recommendation for the starting age for mammography screening from 50 to 40 years[1]. Previously, the Task Force deemed screening in 40-to-50-year-old women a personal choice. Because USPSTF recommendations are so influential, mammography screening for women in their 40s will probably become a health care performance measure; if so, it will effectively become a public health imperative with which primary care practitioners must comply. Such a change will affect more than 20 million U.S. women, and it raises some important questions. First, is there new evidence that mortality from breast cancer is increasing? To the contrary, there has been a steady decrease in breast-cancer mortality in the United States — a major success story of modern medicine. The reduction has been most pronounced among women under 50, whose breast-cancer mortality has been cut in half over the past 30 years, according to the National Vital Statistics System. Similar patterns are seen in other high-income countries, including both those where screening of women in their 40s is very rare (Denmark and the United Kingdom) and those where screening is rare in all age groups (Switzerland) — which suggests that the decline has resulted largely from improved treatment, not screening (see graphs). Second, is there new evidence that the benefit of mammography is increasing? Since the previous USPSTF recommendation was made, there have been no new randomized trials of screening mammography for women in their 40s. Eight randomized trials for this age group, including the most recent (the U.K. Age trial), revealed no significant effect[2]. This finding reflects both the rarity of death related to breast cancer among women in their 40s and the fact that screening reduces mortality less than was hoped — perhaps because more aggressive disease occurs in this age group. Fast-growing cancers are more likely to be missed by screening, often appearing in the interval between exams. Instead of new trial data, the new recommendation is based on statistical models that estimate what might happen if the starting age were lowered. The models assume that screening mammography reduces breast-cancer mortality by about 25% and conclude that screening 1000 women from 40 to 74 years of age, instead of 50 to 74, would result in one to two fewer breast-cancer deaths over a lifetime. The USPSTF's increasing reliance on complex statistical modeling is problematic. Estimated effects can be extremely sensitive to modeling assumptions, which often reflect the conventional wisdom at the time. One prominent model, conducted before the Women's Health Initiative study, projected that nearly all postmenopausal women would have their life expectancy increased by hormone replacement therapy. Models may have the appeal of apparent quantitative precision, but they are only as reliable as their input data and assumptions. As others have argued, policymakers should use models only if they understand the parameters and assumptions underlying them[3]. In this case, it is particularly problematic that the modeled 25% relative risk reduction in breast-cancer mortality with mammographic screening exceeds that observed in meta-analyses of the randomized trials: a 16% relative risk reduction for all eight trials combined (95% confidence interval [CI], 27% to 4% reduction) and a 13% relative risk reduction in the three trials with low risk of bias (95% CI, 27% reduction to 3% increase)[2]. So does the balance of benefits and harms support a new public health imperative? Relative risk reductions can be misleading since they contain no information about absolute risk, which is already low and steadily decreasing for this age group. To clarify the potential effects of the updated guideline in absolute terms, the table summarizes the benefits and harms. For U.S. women in their 40s, the risk of death from any cause in the next 10 years is about 3% regardless of screening. The modeled benefit of mammography is a reduction of a woman's 10-year risk of death from breast cancer from about 0.3% to about 0.2%, a difference of 0.1 percentage point (one breast-cancer death per 1000 women screened for 10 years). In other words, with screening, the likelihood of not dying from breast cancer in the next 10 years increases from 99.7% to 99.8%. This effect is small, particularly in light of the potential harms and what seem to be overly optimistic assumptions of benefits. By far the most common outcomes are false alarms: the USPSTF model estimates that 36% of women 40 to 49 years of age will have at least one in a 10-year course of biennial screening. All will require more testing to prove they don't have cancer; some will undergo multiple tests and face substantial out-of-pocket costs. And some will experience fear: about a third of women describe the experience as “very scary” or “the scariest time of my life”[4]. Some 6.6% of women screened will have a false alarm requiring a biopsy. In addition, the USPSTF model estimates that 0.2% will be overdiagnosed and treated for a cancer not destined to cause harm. These harms may be more frequent in practice, since the model's input from the Breast Cancer Surveillance Consortium probably reflects rates from only high-performing practices. The harms will be more frequent if screening occurs annually rather than biennially, as is the current practice for most U.S. women. The table captures the critical trade-off for women in their 40s: do the benefits, which will accrue to few women, outweigh the harms that will affect many more? The answer is a value judgment that women should be enabled to make for themselves, rather than having a public health imperative imposed on them by physicians with an incentive to meet a “quality” metric. Given the steadily decreasing mortality over the past 30 years attributable to improved treatments, it's likely that fewer and fewer women will benefit from screening over time, while more screening will increase the harms. The Task Force also argues that the new recommendation is an important first step in reducing the disparity between Black and White women in mortality from breast cancer. Although mortality among women in their 40s has fallen roughly by half in both groups since 1990 (see the Supplementary Appendix, available at NEJM.org), the disparity is disturbing and persistent: Black women remain considerably more likely to die from breast cancer than White women (23 vs. 13 deaths per 100,000 women), according to the National Vital Statistics System. But it's hard to imagine how recommending the same intervention to both groups would reduce the disparity, particularly given that screening rates are already similarly high for Black and White women in their 40s (about 60%, according to the National Center for Health Statistics). More screening can't address underlying differences in cancer biology: the incidence of triple-negative breast cancer (which lacks expression of the estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2) among Black women is twice that among White women, according to the National Cancer Institute. This subtype is the most aggressive, has the least effective treatments, and is the most likely to be missed by screening[5]. Nor would earlier screening address the problems facing poor women, who tend to be disproportionately Black, such as the lower quality of medical services available, delayed follow-up on abnormal scans, delays to treatment, and less use of adjuvant therapy. Indeed, lowering the screening age could actually exacerbate the problems contributing to the disparity — by diverting resources toward expanded screening. We need to do more of what really works: ensure that high-quality treatment is more readily accessible to poor women with breast cancer. A change in mammography recommendations would be supported if there were evidence that breast-cancer outcomes were worsening or if there were new evidence that screening younger women had clear benefits. In fact, neither condition applies. We hope that policymakers will reconsider the decision to lower the starting age for mammography screening. The Task Force's models are insufficient to support a new public health imperative, given the limited benefits and such common and important harms to healthy women. It would be better to allow women to make their own decisions based on their own assessment of the data and their values — and to redirect resources to ensuring that all women with breast cancer receive the best and most equitable treatment possible. REFERENCES Preventive Services Task Force. Draft recommendation statement — breast cancer: screening. May 9, 2023. www.uspreventiveservicestaskforce.org/uspstf/draft-recommendation/breast-cancer-screening-adults Gotzsche PC, Jorgensen KJ. Screening for breast cancer with mammography. Cochrane Database Syst Rev. 2013;2013(6):CD001877-CD001877. Kramer BS, Elmore JG. Projecting the benefits and harms of mammography using statistical models: proof or proofiness? J Natl Cancer Inst. 2015;107(7):djv145-djv145. Schwartz LM, Woloshin S, Fowler FJ Jr, Welch HG. Enthusiasm for cancer screening in the United States. JAMA. 2004;291:71-78. Hayse B, Hooley RJ, Killelea BK, Horowitz NR, Chagpar AB, Lannin DR. Breast cancer biology varies by method of detection and may contribute to overdiagnosis. Surgery. 2016;160:454-462.
DOI: 10.1056/NEJMp2307229
|
