Cancer screening recommendations must change as knowledge grows

Page 1 of 6

Posted January 25, 2012

Cancer screening recommendations must change as knowledge grows The onslaught of studies and new updates can hinder informed decisionmaking. Several new sets of cancer screening recommendations, as well as recent studies intended to evaluate the effects of such procedures, have called into question longstanding guidance regarding screening for certain solid tumors. The flood of information has resulted in discordant messages that have caused debate among the public and health care professionals about what is best for patients. Story continues below↓

“The public is not optimally educated. Much like with every medical and scientific area, cancer screening is way more complicated than we would like it to be,” said Donald L. Trump, MD, president and CEO of Roswell Park Cancer Center in Buffalo, N.Y. “It is hard work to communicate the complexity of cancer screening to medical professionals, let alone to individuals who do not do this for a living.” That complexity only increases as more research is conducted into the benefits — and potential harms — of cancer screening. Because cancer screening is not black and white, it is becoming increasingly necessary to start communicating the “grayness” to the public, Trump said. The ACS agrees with that priority. In December, it published a paper in the Journal of the American Medical Association that promised a new process for creating more “trustworthy” cancer screening guidelines. In the paper, Brawley and colleagues said differing guidelines based on the same data have led to system-wide confusion about the value of cancer screening. To combat this, the organization will work to create a more “transparent, consistent and rigorous” process for the development and communication of guidelines. “Ideally, we would like people to understand that screening is not perfect, but that it can save your life by detecting cancer early,” said Robert Smith, PhD, a cancer epidemiologist and director of cancer screening at ACS headquarters in Atlanta. “However, they also have to know that it will not benefit everyone, and that there are downsides associated with screening.” HemOnc Today spoke with several experts about the topic, and they all agreed on one thing: As science progresses and the medical field learns more, screening recommendations are going to be — and should be — updated.

Robert Smith, PhD, a cancer epidemiologist and director of cancer screening at the ACS headquarters in Atlanta, said the public must understand that screening can save your life but is not perfect, and there are downsides associated with it. Photo courtesy of American Cancer Society. Reprinted with permisssion.

Ultimately, many patients will make the decision to screen based primarily on what their physician recommends. That means physicians

http://www.hemonctoday.com/articlePrint.aspx?type=print&rID=91920

2/10/2012

Cancer screening recommendations must change as knowledge grows

Page 2 of 6

must be committed to understanding all of the most recent evidence about screening and take the time to have thorough conversations with their patients. A willing public Most Americans said they believe in the benefits of cancer screening, studies show. A 2004 study in JAMA found that 87% of adults thought routine screening was almost always a good idea, and 74% said finding cancer early saved lives most or all of the time. In addition, respondents showed little concern about false-positive tests that might lead to unnecessary treatment. A study published in Medical Decision Making in 2010 indicated this zeal for screening may be based on a lack of knowledge about it. The study found participants consistently overestimated the risks for being diagnosed with and dying of cancer. Ninety percent of discussions about screening addressed the positive aspects, whereas only 19% to 30% addressed the drawbacks, the researchers said. “The public … has been inundated by oversimplified messages about screening,” said Barnett S. Kramer, MD, MPH, director of the NCI’s Division of Cancer Prevention. “They receive sound bites that emphasize the benefits, and the public is not fully aware of the harms of screening.” A commentary by Michael Edward Stefanek, PhD, published in November in the Journal of the National Cancer Institute, suggested a shift in the health care community’s approach to cancer screening from developing and debating guidelines to educating and informing the public. The “long history of in-fighting” degrades the public’s confidence in health care professionals, according to Stefanek. However, Kramer said he doubts a simple solution exists. “Different groups are going to look at the same evidence and come to different conclusions,” Kramer said. “In the meantime, the public is going to be subjected to competing messages.” Gynecologic cancers Maurie Markman, MD, a medical oncologist and HemOnc Today Editorial Board member, works in an area of oncology in which screening recommendations and the benefits of screening would seem to be relatively clear cut.

Donald L. Trump

In cervical cancer, widespread screening was introduced with the Pap smear, which — if done regularly — has been shown to prevent most cervical cancers. In ovarian cancer, the opposite is true. Results from the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial presented at the 2011 ASCO Annual Meeting and subsequently published in the Journal of the American Medical Association showed that the use of CA-125 and transvaginal ultrasound conferred no benefit in terms of risk for mortality compared with usual care. They also were associated with an increased risk for false-positives and unnecessary invasive procedures. “There is currently no evidence whatsoever for the value of screening in ovarian cancer. In fact, [results from PLCO] showed very clearly that there is a potential for harm,” said Markman, senior vice president of clinical affairs and national director for medical oncology at Cancer Treatment Centers of America. However, that does not mean there is system-wide consensus. Research continues into the use of Pap smears vs. HPV tests for cervical screening, as well as the frequency at which women should undergo these tests. In ovarian cancer, a recently published small, nonrandomized study from the University of Kentucky — which found that annual ultrasonographic screening attained increased detection of epithelial ovarian cancer — further confused the issue, even though the study was not designed to establish whether the increased detection translates into a reduced risk of dying of ovarian cancer. “There is currently no evidence that screening for ovarian cancer is of value, but that does not mean that the public is getting that message,” Markman said. “Women who are concerned [about ovarian cancer] walk into the gynecologist or general practitioner every day and ask for CA-125 or ultrasound.” That is because, for years, the public has been told repeatedly that early detection of cancer is critical. “The reason for that is that earlier-stage disease has a better prognosis, tumor after tumor,” Markman said. “It is very hard to argue with that logic, except that the data does not always support it.” Lung cancer

http://www.hemonctoday.com/articlePrint.aspx?type=print&rID=91920

2/10/2012

Cancer screening recommendations must change as knowledge grows

Page 3 of 6

Breakthroughs on an effective screening method for lung cancer, the leading cause of cancer death in the United States, have been anticipated for years. In August, the NCI published the results of its National Lung Cancer Screening Trial (NLST), which compared CT scans to chest X-rays in more than 53,000 former and current heavy smokers aged 55 to 74 years. Five-year results published in The New England Journal of Medicine indicated there was a 20% relative reduction in deaths from lung cancer among patients screened annually (at baseline and for two subsequent years) with CT compared with chest X-ray.

Barnett S. Two months later, the results of the PLCO lung cancer arm were published in JAMA. In Kramer the lung cancer arm of the trial, more than 150,000 patients were randomly assigned to either annual screening with chest radiograph or usual care. Researchers did not observe a statistically significant difference in lung cancer mortality between the two groups, although these findings are less relevant now that the NLST has demonstrated that screening for lung cancer should be done with CT and not chest X-ray. “The NLST told us that if you screen a particular group of patients three times annually that you prevent one in five deaths from lung cancer compared with X-ray,” said Peter B. Bach, MD, director of the Center for Health Policy and Outcomes at Memorial Sloan-Kettering Cancer Center. However, these results should raise legitimate concerns that the one-in-five reduction in death rate achieved in NLST was the best that can be achieved, Bach said. “The centers where this study was done had a lot of expertise and the compliance rate was incredibly high,” he said. “If this was unleashed more broadly, there could be a real fall-off in efficacy.” The NCCN recommends low-dose CT screening for select patients at high risk for lung cancer, but no other major association or society currently recommends routine lung cancer screening. Bach said he feels that the value of lung cancer screening is still “up in the air,” adding that the field still needs extensive research. Breast and prostate cancers Breast cancer and prostate cancer are the two areas where most of the confusion and conflicting information about screening seem to arise, and much of it revolves around recommendations issued by the US Preventive Services Task Force (USPSTF). In 2009, the group issued a recommendation against routine mammography in asymptomatic women aged 40 to 49 years. It also recommended that the frequency of mammograms be reduced to every 2 years in women aged 50 to 74 years. This past fall, the Canadian Task Force on Preventive Health Care echoed the USPSTF’s stance, recommending against routine mammograms in women aged 40 to 49 years. Within weeks of this recommendation, the public was blitzed again with media reports about studies that examined the usefulness of mammograms. Some headlines claimed mammograms reduced breast cancer deaths by 50%, whereas others suggested mammograms caused more harm than good. Anyone who cuts through the noise will realize the message to the public has not changed much, Smith said. “The general consensus is that regular breast cancer screening is an important part of a Peter B. Bach preventive health plan,” he said. “Although there are some differences in the agespecific recommendation, the overall guidelines are more similar than they are different.” A recommendation, after all, is just a recommendation. Women in their 40s should discuss their desire to screen for breast cancer — or any other type of cancer — with their doctors, Smith said. More recently, the USPSTF issued a draft recommendation against the routine use of PSA testing in healthy men. This caused a conflict among certain medical experts, renewing the debate about the usefulness of prostate cancer screening vs. the potential negative effects. “Despite this recent flurry of activity and the task force recommendations, I believe that the majority of practitioners are still following the same routine,” said Trump, HemOnc Today’s genitourinary cancer section editor. “For men 50 years or older with average risk, they get a PSA and [digital rectal examination]. That same approach starts at age 40 for high-risk individuals.”

http://www.hemonctoday.com/articlePrint.aspx?type=print&rID=91920

2/10/2012

Cancer screening recommendations must change as knowledge grows

Page 4 of 6

However, Trump said he recognizes that the issue of cancer screening — particularly in prostate cancer — is complex, and none of the studies examining the topic are perfect. “I frequently hear patients or friends say that medical science does not know what they are doing, that they keep changing recommendations or can’t get anything straight,” he said. “I absolutely still recognize the importance of cancer screening, but as it gets more complex and less clear cut, it is going to mean longer discussions between physicians and patients.” A difficult adjustment The debate about cancer screening is not new. In fact, a certain degree of debate ensues after any change in screening recommendations because, although studies and research involve trends and logic, cancer screening — and the prospect of helping to save someone’s life — also involves emotion. A study published in the Archives of Internal Medicine found that, despite the USPSTF recommendation against regular screening in adults aged 75 years or older, more than half of study participants in that demographic recently had been screened for breast (57%), cervical (62%), colorectal (53%) or prostate cancers (56%). That is because, on the whole, there is still considerable support for cancer screening among those in the health care and oncology community. However, as more research is done in the areas of solid tumor cancer screening, researchers may discover screening does not provide benefits in every area. That realization may be difficult for the public and health care professionals to understand and apply. “Habits are hard to break, and for most things that physicians do, there is evidence and then there is belief,” Bach said. “For practitioners, screening is definitely one of those places.”

Maurie Markman

Markman agreed, adding that resistance to change may not be a question of unwillingness to listen to new evidence, but rather a reluctance to buy into statistical evidence over what they view as first-hand evidence. “A physician may have screened a patient for a particular cancer … and found cancer in that patient,” Markman said. “To them, that shows that screening works, and understandably so. So much of being a doctor is learning from individualized experiences over time.”

However, it is important that physicians resist the urge to impose their value system on a patient, Kramer said. “Physicians and patients need to know what the evidence is, and they need to be acquainted with the harms and benefits,” he said. “It is important for the target population, which is generally healthy people, to become informed when they make the decision to be screened, because we are learning more and more that sound bites simply overstate the evidence.” – by Leah Lawrence Disclosure: Drs. Bach, Kramer, Markman, Smith and Trump report no relevant financial disclosures.

For more information:

http://www.hemonctoday.com/articlePrint.aspx?type=print&rID=91920

2/10/2012

Cancer screening recommendations must change as knowledge grows

Page 5 of 6

Bellizzi KM. Arch Intern Med. 2011;171:2031-2037. Brawley O. JAMA. 2011;306:2495-2499. Buys SS. JAMA. 2011;305:2295-2303. Canadian Task Force on Preventive Health Care. Screening for Breast Cancer. Available at: www.canadiantaskforce.ca/recommendations/2011_01_eng.html. Accessed Jan. 6, 2012. Hoffman RM. Med Dec Making. 2010;3:53S-64S. Oken MM. JAMA. 2011;306:1865-1873. Schwartz LM. JAMA. 2004;291:71-78. Stefanek ME. J Natl Cancer Inst. 2011;doi:10.1093/jnci/djr474. The National Lung Screening Trial Research Team. N Engl J Med. 2011;365:395-409. van Nagell JR Jr. Obstet Gynecol. 2011;118:1212-1221.

Is mortality the appropriate endpoint for clinical trials examining the benefit of cancer screening?

All-cause mortality is an ideal endpoint, but it is not feasible. In an important paper, Black and colleagues identified two problems with the use of a cancer-specific mortality endpoint in a cancer screening trial. First, deaths from an uncertain cause are more likely to be incorrectly attributed to cancer if there was a previous diagnosis of cancer, especially if the diagnosis was relatively recent. Black called this “stickydiagnosis” bias. Sticky-diagnosis bias translates into a higher cancer death rate in the intervention group than would be the case without it. Second, deaths that are caused or triggered by screening, workup or a subsequent therapy (eg, perforation of the colon and, perhaps, cardiovascular deaths) are less likely to be correctly attributed to screening. Black called this “slippery-linkage” bias. Slippery-linkage bias translates into a lower cancer death in the intervention group than would be the case without it. But it is unlikely these two opposing effects would balance, so there is cause for concern. Thus, if cost were not a constraint, the ideal endpoint of a cancer screening trial would be all-cause mortality. However, this ideal endpoint is not achievable in practice. Based on calculations in Baker et al, a cancer screening trial with a cancer-specific mortality endpoint that would involve 150,000 participants would require 4.1 million participants for an all-cause mortality endpoint, and that would unlikely be funded.

Stuart Baker

To minimize sticky-diagnosis and slippery-linkage bias, Baker et al recommended (1) using cancer death as an endpoint, with careful review of the death records, and (2) counting as a cancer death from screening any noncancer deaths attributable to screening or treatment for the cancer. In addition, Baker et al recommended all deaths and their causes be reported. They also recommended determining if there is a significantly different estimated probability of a particular noncancer cause of death after adjusting for multiple comparisons. If so, investigators should re-examine the death records to check for biases. If there are no obvious biases, the investigators should consider the possibility that screening or treatment was responsible for this difference and pursue a more detailed review. Stuart Baker, ScD, is a mathematical statistician with the Division of Cancer Prevention at the NCI. References: Baker SG. BMC Med Res Methodol. 2002;2:11. Black WC. J Natl Cancer Inst. 2002;94:167163. Disclosure: Dr. Baker reports no relevant financial disclosures.

Cancer-specific mortality is best for today, but maybe not good enough for tomorrow.

http://www.hemonctoday.com/articlePrint.aspx?type=print&rID=91920

2/10/2012

Cancer screening recommendations must change as knowledge grows

Page 6 of 6

I agree with the views of Dr. Baker about the proper mortality metrics in cancer screening trials. I offer this counterpoint not to disagree, but more to point out that our agreement reflects only our shared past experience. More effective cancer therapies and new screening methods will bring new challenges for measuring the impact of cancer screening trials of the future. Dr. Baker reminds us of a couple of tactile-rich metaphors described by Black et al: a “sticky-diagnosis” and a “slippery-linkage” bias. To these, I would add two important “slippery slope” assumptions. The first slippery slope assumption is the inherent belief we have formed from our past experience that, once cancer progresses far enough, death becomes inevitable enough to be a good outcome measure. Now imagine a future in which we have many more effective treatments for late-stage cancers, where the slope toward death becomes much stickier. For some cancers, that future is now. We already have the ability to extend life of stage IV breast cancer patients by many more years than in the era when the first mammography trials were completed. Any mammography trial being designed today should consider the avoidance of morbidities and costs from treating advanced-stage disease as meaningful outcomes, not only death from breast cancer. The second slippery slope is the assumption that there is only one slope to be concerned with after cancer screening. Indeed, it is now clear that overdiagnosis also is a slippery slope of its own that causes some screened patients to unnecessarily experience the morbidities and costs of cancer diagnosis and treatment.

Tim Byers

Now imagine a future where cancer screening tests become much more sensitive for extremely early disease, like advanced imaging tests for primary screening or detecting cancer cells or bits of cancerspecific genetic sequences in the circulation. These ultra-sensitive cancer screening tests inevitably will increase the problem of overdiagnosis, so trials assessing them will need to better measure the various harms people experience on that slippery slope. In the future, better cancer treatment will extend life much longer for patients with advanced disease, and advanced imaging and biomarker screening tests will identify cancers much earlier. Therefore, in future cancer screening trials, cancer-specific mortality endpoints will need to share more of the spotlight with many other types of outcome measures, including the morbidities and costs resulting from cancer treatment and overdiagnosis. Tim Byers, MD, MPH, is associate dean for public health practice at Colorado School of Public Health and associate director for cancer prevention and control at University of Colorado Cancer Center in Aurora, Colo. Disclosure: Dr. Byers reports no relevant financial disclosures. Copyright © 2012 HemOnc Today. All rights reserved.

http://www.hemonctoday.com/articlePrint.aspx?type=print&rID=91920

2/10/2012