肺癌CT筛查每年可避免1.2万美国人死亡

最新研究显示，如果美国对全国健康吸烟者进行肺癌CT筛查，每年可推迟或避免12,250例死亡，这相当于美国所有肺癌死亡的7.6%(Cancer 2013 Feb. 25 [doi: 10.1002/cncr.27813])。

本研究由美国癌症学会监测研究方面的高级流行病学家Jiemin Ma博士及其同事进行，采用最新的全国健康调查和2010年美国人口普查数据，并采用全国肺筛查试验(NLST)的患者标准：患者年龄55~74岁，每年吸烟至少30包，当前仍吸烟或过去15年已戒烟。

NLST试验显示，与胸部X射线检查相比，小剂量CT(LDCT)检查在6.5年内使肺癌死亡率降低了20%(N. Engl. J. Med. 2011;365:395-409)。NLST试验估计约有7百万美国人适合接受LDCT筛查，而本研究作者估计这一数字为860万。一些因素可能影响了新的估计数字，包括使用NLST报告的筛查效应作为参数来确定可避免肺癌死亡的人数。作者表示，尽管直接使用NLST筛查效应的方法具有简单易懂的优点，但这也使本研究的估计数字只能在NLST试验设计和筛查方案的框架内解读。NLST试验未在每年吸烟少于30包或吸烟年龄较早的患者中探讨筛查的有效性，LDCT筛查组20%的死亡率降幅可能被低估。Ma 博士还指出，NLST试验中，对照组使用的是胸部X射线(CXR)，这种情况不可能发生在一般人群。如果CXR也能够有效预防肺癌死亡的话，那么这可能也导致可避免的肺癌死亡例数被低估。其他可能导致死亡率被低估的因素包括使用自报的吸烟数据，这一数据不太可靠，因为吸烟者报告的烟草使用量往往比实际低。另一方面，基于2000年至20006年的死亡例数估计了适合筛查人群的肺癌死亡率，这可能高估了目前的死亡率。研究模型还假设，将有100%的目标人群接受筛查，这在现实中不可能达到。在筛查率为100%和肺癌死亡率降低30%的理想情况下，LDCT筛查每年可避免18,375例肺癌死亡。但如果860万合格人群中每年仅有70%接受筛查的话，可避免的肺癌死亡例数就降至8,575例。

在随刊述评中，西雅图华盛顿大学卫生服务中心主任Larry Kessler博士指出，在NLST试验中，24.2%的LDCT筛查结果为阳性，在这些阳性结果中，96.4%为假阳性结果。本研究报告的最新估计数字虽然非常重要，但研究者采用的方法存在缺陷。本研究报告的是1年估计值而非更常见的拯救生命/年值，后者能够反映随时间推移的筛查效应。本研究可能低估了全国CT筛查项目随时间推移的总体效应。

此外，本研究未探讨各种影响吸烟的因素，如年龄、月经和队列效应。目前吸烟的女性人数出现惊人增加，这将改变肺癌死亡的性别差异。在本研究适合筛查的520万男性和340万女性中，LDCT筛查将可避免8,990例男性死亡和3,260例女性死亡。

本研究获得了美国癌症学会内部研究部门的支持。Ma博士及其同事和Kessler博士均声明无经济利益冲突。

爱思唯尔版权所有  未经授权请勿转载

By: PATRICE WENDLING, Oncology Practice

Roughly 12,250 deaths could be delayed or prevented each year if a national lung cancer computed tomography screening program were implemented among heavy smokers in the United States.

That represents about 7.6% of all lung cancer deaths in the nation, according to a new study led by Jiemin Ma, Ph.D., a senior epidemiologist in surveillance research with the American Cancer Society (Cancer 2013 Feb. 25 [doi: 10.1002/cncr.27813]).

To make the calculation, the authors used the most recently available National Health Interview Survey and 2010 U.S. Census data, and applied the National Lung Screening Trial (NLST) criteria of patients aged 55-74 years with at least 30 pack-years of smoking who were still smoking or had quit in the last 15 years.

The landmark NLST demonstrated a 20% reduction in lung cancer mortality over 6.5 years using low-dose computed tomography (LDCT) screening versus chest radiography (N. Engl. J. Med. 2011;365:395-409). That study sparked debate about how – or even if – a national lung cancer screening program should be implemented.

The authors of the current study estimated that 8.6 million Americans would be eligible for LDCT screening. That’s up from 7 million identified in the NLST.

Several factors might have affected the new estimates, including use of the NLST-reported screening effect as a parameter in determining the number of avertable lung cancer deaths.

Although this approach is "straightforward and easy to understand," direct use of the NLST screening effect "confines our estimates to be interpreted within the context of the NLST study design and screening protocol," the authors wrote.

The NLST did not address the effectiveness of screening in patients with fewer pack-years or starting at an earlier age, and the study’s investigators noted that the 20% mortality reduction with LDCT screening may be an underestimated effect.

Dr. Ma also pointed out that the NLST used chest x-ray (CXR) for its control group, something unlikely to occur in the general population. This potentially also could lead to an underestimation of avertable lung cancer deaths "if CXR was of benefit in preventing lung cancer mortality," he said.

Other factors that may have led to underestimation include the use of self-reported smoking data, which can be a bit unreliable because smokers tend to underreport their tobacco use.

On the other hand, the lung cancer mortality rates for screening-eligible populations were estimated based on deaths occurring between 2000 and 2006, possibly overestimating the current rates, the authors observed.

The model also assumed that 100% of the target population would undergo screening, which is unlikely to be achieved in the real world.

Under an optimal scenario of 100% screening uptake and a 30% reduction in lung cancer death rates, 18,375 lung cancer deaths would be averted each year by LDCT screening, Dr. Ma reported. If just 70% of the 8.6 million eligible people are screened annually, however, the number of lung cancer deaths averted falls to 8,575.

"Ultimately, the magnitude of potential benefit from LDCT screening in the community remains to be determined," he wrote.

The American Cancer Society and American Lung Association back screening for heavy smokers based on the NLST criteria, while National Comprehensive Cancer Network (NCCN) surveillance guidelines are somewhat broader and include smokers aged 50 years or older with a smoking history of 20 pack-years plus an additional lung cancer risk factor.

In an accompanying editorial titled "Is 20% of a Loaf Enough?" Larry Kessler, Sc.D., cautioned against "haphazard adoption" of LDCT outside of organized and clear screening protocols, citing such potential harms as inconsistent follow-up with providers and the fear generated by false positives (Cancer 2013 Feb. 25 [doi: 10.1002/cncr.27811]). In NLST, 24.2% of LDCT scans were positive, and 96.4% of those positive results were false-positive findings, noted Dr. Kessler, chair of health services at the University of Washington in Seattle.

While the new national "estimates are important," the approach taken by the investigators "missed some critical elements of population dynamics that may alter what we would really experience with a nationally organized screening program," he cautioned.

First, the study reported 1-year estimates instead of the more common life-years saved approach, which is necessary to appreciate the effect of the screening program over time.

"The likely effect is that Ma et al. may have underestimated the total effect over time of a national CT screening program," Dr. Kessler said.

He also suggested the analysis failed to address the dynamics of smoking, which are influenced by age, period, and cohort effects – and must be tied to estimating the long-term effect of a screening program. An obvious example of that is the alarming increase in smoking among women, Dr. Kessler noted, which would alter the discrepancy in lung cancer deaths between the sexes, because the smoking histories of women ultimately "catch up."

Of the 5.2 million men and 3.4 million women eligible for screening, an estimated 8,990 deaths in men and 3,260 deaths in women would be averted with LDCT screening, according to the study’s analysis.

The study was supported by the Intramural Research Department of the American Cancer Society. Dr. Ma, his coauthors, and Dr. Kessler made no disclosures.

View on the NewsMore at-risk patients could be identified

Comments from Dr. Martin C. Tammemägi: Ma et al. present the annual number of deaths delayed or prevented by low-dose computed tomography lung cancer screening. Even if the overall estimate of 12,000 may be inaccurate for a number of reasons, the overall magnitude is large enough to serve as a slap in the face. The editorial by Dr. Kessler very nicely discusses many relevant points regarding lung screening. For emphasis, I reiterate some of Dr. Ma’s and Dr. Kessler’s points.

Dr. Ma and his colleagues based their primary results on the NLST-observed lung cancer mortality reduction of 20% and application of NLST entry criteria for selection of screenees. This mortality reduction was observed in the ideal randomized screening trial setting.

In everyday practice, screening success may fall far well short of the trial-observed mortality. Widespread public health screening initiatives work best if implemented as systematic programs rather than “opportunistic” screening, as is most likely to occur in the United States. Thus, achievement of 20% lung cancer mortality reduction may be difficult to obtain.

On the other hand, long-term annual screening may lead to greater effectiveness of screening than was observed in the NLST with only three screens. Furthermore, if one selected 8.6 million people for screening based on elevated lung cancer risk prediction model probabilities, an additional 2,750 deaths may be averted, as an estimated 41% fewer lung cancer would be missed, compared with application of the NLST criteria (N. Engl. J. Med. 2013;368:728-36).

Whether to and how to optimally implement lung cancer screening is still muddled. Within the next year, we anticipate several important additions to lung cancer screening knowledge: CISNET models will be published that will shed light on the impact of varying screening parameters; the U.S. Preventive Services Task Force is expected to provide guidance with regard to lung screening; and pulmonary nodule malignancy prediction models based on prospective population-based data will help guide clinicians with respect to true- vs. false-positive lung screens.

Dr. Tammemägi is a professor of epidemiology at Brock University in St. Catharines, Ont.