Abstract
Several important lessons have been learnt from our experiences in screening for various cancers. Screening programmes for cervical and colorectal cancers have had the greatest success, probably because these cancers are relatively homogenous, slow-growing, and have identifiable precursors that can be detected and removed; however, identifying the true obligate precursors of invasive disease remains a challenge. With regard to screening for breast cancer and for prostate cancer, which focus on early detection of invasive cancer, preferential detection of slower-growing, localized cancers has occurred, which has led to concerns about overdiagnosis and overtreatment; programmes for early detection of invasive lung cancers are emerging, and have faced similar challenges. A crucial consideration in screening for breast, prostate, and lung cancers is their remarkable phenotypic heterogeneity, ranging from indolent to highly aggressive. Efforts have been made to address the limitations of cancer-screening programmes, providing an opportunity for cross-disciplinary learning and further advancement of the science. Current innovations are aimed at identifying the individuals who are most likely to benefit from screening, increasing the yield of consequential cancers on screening and biopsy, and using molecular tests to improve our understanding of disease biology and to tailor treatment. We discuss each of these concepts and outline a dynamic framework for continuous improvements in the field of cancer screening.
The proximate goal of cancer screening is the identification of early stage cancer, or precancerous lesions, before a person develops symptoms and at a point in the disease trajectory when treatment is likely to result in cure. This concept is simple, but practicing effective screening on a population level is a complex endeavour. In 1968, Wilson and Jungner1 of the WHO proposed criteria that should be met before a screening test should be implemented (BOX 1); these principles continue to guide policy in countries where implementation of organized screening programmes is being considered. For a number of common cancers, some of these criteria have been met; however, many continue to present challenges and remain incompletely addressed (BOX 1). Wilson and Jungner’s suggestion that “the natural history of the condition, including development from latent to declared disease, should be adequately understood” (REF. 1) seems particularly prophetic. At the time of the WHO report, and for decades after, the prevailing model of carcinogenesis was that of a linear progression from precursor disease to early stage (localized) cancer and, subsequently, to advancedstage (disseminated) cancer. Indeed, the models of colorectal cancer (CRC) tumorigenesis proposed by Vogelstein et al.2 in the late1980s suggested a relatively slow, stereotyped evolution from colonic polyp to cancer, commensurate with the acquisition of certain mutations over time. A similar paradigm has become established for the natural history of cervical cancer, and healthcare organizations in a number of countries, including the USA, introduced screening for breast and prostate cancers, presuming that these diseases also followed this classic developmental framework.
Box 1 |. Cancer screening in 2016: meeting the Wilson and Jungner1 criteria?
- The condition sought should be an important health problem
- Criterion met
- There should be an accepted treatment for patients with recognized disease
- Criterion met
- Facilities for diagnosis and treatment should be available
- Criterion met
- There should be a recognizable latent or early symptomatic stage
- Criterion not fully met. Owing to the spectrum of disease heterogeneity, more often true for some cancer types (cervical and colorectal), but less often true for other types (breast, prostate, and lung)
- There should be a suitable test or examination
- Criterion met
- The test should be acceptable to the population
- Criterion met
- The natural history of the condition, including development from latent to declared disease, should be adequately understood
- Criterion not fully met. Focus for improvement: cervical intraepithelial neoplasia, ductal carcinoma in situ, colonic polyps, lung nodules, and indolent invasive cancers (for example, Gleason 6 prostate cancers)
- There should be an agreed policy on whom to treat as patients
- Criterion not fully met. Focus for improvement: management of disease entities listed in above
- The cost of case-finding (including diagnosis and treatment of patients diagnosed) should be economically balanced in relation to possible expenditure on medical care as a whole
- Criterion not fully met. Focus for improvement: refining targets of screening and biopsy to improve yield and focus on precursor or early stage forms of potentially morbid disease
- Case-finding should be a continuing process and not a “once and for all” project
- Criterion not fully met. Focus for improvement: screening registries should be established to facilitate quality improvement
With mass implementation of screening for cancer, our experiences on the population level have deepened our understanding of cancer biology. Screening efforts have revealed a previously unappreciated reservoir of precancerous lesions and indolent cancers that would not have otherwise come to clinical attention. By contrast, other cancers have been recognized to grow so fast that screening assessments performed at predetermined intervals do not enable detection before their spread to local or distant organs. Indeed, we now understand that ‘cancer’ comprises a heterogeneous collection of diseases, both across and within organ sites. The advent of geneexpression profiling and other molecular diagnostic methodologies has advanced our understanding of cancer biology beyond the original model proposed by Vogelstein and colleagues. In fact, treatment decisions are increasingly being guided by geneexpression profiling, rather than by traditional factors, such as disease stage or histopathological features3.
The challenge in screening for and prevention of disease relates to the concept that it is difficult to make healthy people better off than they already are, but not as difficult to make them worse off. Screening, by virtue of increasing the likelihood of performing a biopsy, will potentially uncover a reservoir of biologically moreindolent cancers, some of which might lack the potential to progress to metastatic disease (the ultimate cause of most cancerrelated deaths). Detection of indolent lesions is not intrinsically harmful, but can lead to downstream diagnostic and therapeutic interventions that cause serious adverse effects to patients. Nevertheless, screening can be of benefit when diagnosis and treatment of a precancerous lesion or an early stage tumour will avert progression of disease to metastasis and/or death. This hope continues to form the basis for population screening for cancer, but also fuels the hype that surrounds cancer screening.
Going forward, lessons learned from the careful distillation of several decades of experience in cancer screening can guide practice and drive improvements in cancer screening. Four key lessons and their corollaries form the foundations for this Review of screening for breast, prostate, cervical, colorectal, and lung cancers (BOX 2). These concepts serve to refine — rather than replace — the Wilson and Jungner criteria, by highlighting the corresponding action points that must be considered to continue improving the delivery of screening assessments. We present a framework for improving cancer screening, based on a stepwise examination of the decisions that must be made before, during, and after deployment of a screening test. Owing to the scope of this topic, emerging technological advancements in screening tests are discussed where relevant, but are not otherwise comprehensively covered.
Box 2 |. Key lessons surrounding cancer screening and their corollaries.
Lesson 1: The biology of invasive cancers ranges from indolent to aggressive
Corollary: Screening will be of greatest benefit if targeted at detecting progressive, potentially morbid disease while avoiding identification, and/or reflexive treatment, of indolent disease
Lesson 2: Not all precancerous lesions are obligate precursors of invasive cancers; in fact, most are not
Corollary: Treatment of precancerous lesions is of greatest benefit when it prevents potentially morbid disease, or otherwise removes precursors of less-aggressive disease in an effective, nontoxic way
Lesson 3a: Effective screening and removal of early stage cancers should cause a concomitant decline in the incidence of advanced-stage cancers
Lesson 3b: Effective screening and removal of precursor lesions should cause a concomitant decline in the incidence of invasive cancers
Corollary: Population-level trends can be analysed to identify unintended consequences of screening, such as overdiagnosis, and drive efforts aimed at improving outcomes
Lesson 4: Not all individuals will benefit equally from screening
Corollary: Screening should be offered to a carefully defined target population after consideration of risk factors and overall prognosis
Screening: a population-based view
Cancer screening can contribute to decreasing cancer morbidity and mortality through two mechanisms: the detection of a precursor lesion, or the early detection of invasive cancer. The benefits of screening are greater when the detection of disease at an earlier (or precancerous) stage improves outcomes; therefore, the available treatment should be safe, acceptable, and more effective when implemented earlier in the disease course.
The identification of true precursor lesions through population screening should result in a decrease in the incidence rates of invasive cancer over time. Colonoscopy and colposcopy (following cervical cytology) enable direct visualization of the target organs (rectum and colon, and cervix, respectively), and concurrent or subsequent removal of atrisk tissue. The use of these approaches depletes the reservoir of precancerous lesions, namely colonic polyps and cervical intraepithelial neoplasia (CIN), which has led to a decrease in the overall incidence of the respective invasive cancers4 (FIG. 1). The success of populationbased screening programmes using cervical cytology in reducing the incidence and mortality rates of invasive cervical cancer fuelled enthusiasm surrounding screening for other (pre)cancers. The detection and removal of all suspected precursor lesions, however, does not lead to the same result in all screening programmes. As is discussed herein, widespread use of mammography screening has increased the frequency of intervention to remove in situ breast lesions, but has not resulted in a decline in the incidence of invasive breast cancer5,6. The underlying biology and heterogeneity of cancers largely determine the tradeoff between the benefits and the harms of screening.
Figure 1. Age-adjusted incidence rates of invasive cancers for which population-based screening is practiced in the USA.
Annual incidence rates in men (for prostate and colorectal cancers) and women (for cervical and uterine, breast and colorectal cancers) over the age of 50 years are shown for a 37‑year period (1975–2012), based on data from the Surveillance, Epidemiology, and End Results (SEER) registry4. Approximate eras of widespread use of the respective screening tests are represented by black lines, with dotted regions representing initial periods of increasing dissemination of the tests following their introduction. The incidence rates of cervical cancer in women and colorectal cancers in both men and women have declined since the early‑to‑mid 1980s, probably owing to the screening-based detection and subsequent removal of cervical intraepithelial neoplasia and colonic polyps, respectively. On the other hand, the incidence rates of prostate cancer and breast cancer have increased over the same timeframe, probably owing to increased detection of localized cancers as a result of the widespread use of prostate-specific antigen (PSA)-based and mammography screening, respectively.
Differences in disease biology between cancers of the same organ site are of particular importance for tests aimed at the early detection of invasive cancer. Such tests rely on either radiographic imaging of a target organ (for example, mammography for breast cancer and lowdose computed tomography (LDCT) for lung cancer), or measurement of a circulating biomarker associated with presence of the disease (for instance, PSA testing for prostate cancer). These tests are beneficial when they detect invasive cancer at an early, localized stage. The desired effect is a ‘stage shift’, whereby the proportion of patients diagnosed with early stage disease increases over time, accompanied by a decline in incidence of advancedstage disease — reflecting averted progression of cancers via early detection and treatment. Importantly, the absolute decrease in the incidence rate of advancedstage disease should be considered, rather than the change in the relative proportions of these cancers versus earlystage disease, as the latter comparison can be falsely reassuring if an excess of early stage cancers that would not otherwise progress to advanced stages is detected through screening7. Additionally, one must consider whether the stage shift is associated with an improvement in diseaserelated mortality, or because this measure is also affected by the efficacy of treatment, the incidence of metastatic cancers8.
The focusing of screening programmes on the early detection of invasive cancer arose from an incomplete understanding of the heterogeneity in cancer biology. Cancers can have a spectrum of clinical behaviours, ranging from indolent to aggressive. At one end of this spectrum lies a subset of cancers so aggressive that screening will not, ultimately, be of benefit. This subset comprises cancers that are prone to early systemic spread and, therefore, have a poor prognosis8. Despite routine screening, patients with these cancers will already have distant metastatic disease at the time of detection. The term ‘interval cancer’ is commonly applied to symptomatic tumours that arise in between screening intervals. These cancers tend to be more aggressive and are diagnosed at moreadvanced stages than screendetected lesions9. Representing more of a limitation of screening, rather than a harm, patients with interval cancers present with clinical symptoms, and at the same disease stages, regardless of screening. Moreover, in clinical studies, these clinically detected cancers are associated with a worse prognosis than those detected as a result of screening10, thus challenging the paradigm that screening is effective at improving patient outcomes for all tumour phenotypes.
Screening predominantly detects lesions other than interval cancers, which necessarily include tumours with slow and moderate growth velocities. A difficult challenge, therefore, is to avoid preferential detection of indolent (slowgrowing) cancers that might not otherwise come to clinical attention; detection of these cancers might increase the incidence of early stage cancers, but is unlikely to substantially reduce the incidence of advancedstage cancers because they would probably never progress to such a stage during the patient’s lifetime. Herein lies a potential harm of screening: in addition to the intrinsic risk of falsenegative and falsepositive results owing to the imperfect sensitivity and specificity of the screening tests, screening incurs ‘overdiagnosis’, defined as the detection of cancerous lesions that would not have caused morbidity or mortality. A closely related concept is ‘overdetection’ — the detection of premalignant lesions that are not destined to progress to malignancy. Patients with premalignant lesions and indolent cancers can be subjected to invasive tests and treatments, or toxic therapies; therefore, the theoretical risks of overdetection can be similar to those of overdiagnosis: ‘overtreatment’. Overtreatment refers to therapy that is inappropriately invasive or extensive in relation to the biology of disease and can occur with a variety of diseases.
Overdiagnosis has been observed on the population level since the 1990s, when screening of children for neuroblastoma was associated with this effect11; however, a particularly illustrative example is that of thyroidcancer screening in the Republic of Korea (South Korea). Widespread governmentsponsored screening in South Korea led to a fivefold increase in the incidence of papillary thyroid cancers without a concomitant decrease in diseasespecific mortality12. Organized populationscreening for thyroid cancer does not exist in the USA, although the incidence rate of thyroid cancer is increasing most rapidly of all cancers, owing largely to opportunistic ultrasonography screening13,14.
The uncovering of a large reservoir of indolent thyroid cancers illustrates the potential for overdiagnosis when screening is targeted at cancer types with a large reservoir of nonprogressive disease (BOX 2: Lesson 1). Similarly, not all precancerous lesions are obligate precursors of invasive disease (BOX 2: Lesson 2). As will be explained in the following sections, populationwide trends, such as those seen for thyroid cancer in the Republic of Korea, can provide valuable clues as to whether screening is having unintended consequences (BOX 2: Lesson 3). In these instances, screening exposes a large population of healthy people to unnecessary harms (BOX 2: Lesson 4). Specifically, overdiagnosis leads to subsequent diagnostic and therapeutic interventions that carry risks, but are ultimately of limited or no benefit (overtreatment).
Thus, screening is likely to be of limited benefit at either extreme of cancer aggressiveness. The challenge is to leverage the experience with screening on the population level gained to date, to continue advancing our understanding of cancer biology, in order to avoid overdiagnosis and overtreatment. In the following sections, we review the two major populationbased screening strategies, detection of precursor lesions and early detection of invasive cancer, to further illustrate the lessons and corollaries outlined in BOX 2.
Detection of precursor lesions
Cervicalcancer screening was adopted based largely on the results of early observational studies that showed a decrease in incidence of the disease coincident with widespread screening15,16. Randomized clinical trials (RCTs) performed in India subsequently revealed a mortality benefit of cervicalcancerscreening programmes17–19. Moreover, high usage of cytologybased screening in US women has been accompanied by a decline in cervicalcancer incidence and mortality (FIG. 1). The causal link between screening and reduced cervicalcancer mortality is also supported by the observation that over half of the incident cervical cancer cases reported each year in the USA and other countries occur in the relatively small subpopulation of unscreened women20,21. Of note, cervicalcancer risk can be entirely eliminated among women who undergo total hysterectomy; the high prevalence of hysterectomy by the age of 65 years among women in the USA — up to 50% — has contributed heavily to the observed low rates of cervical cancer in this population22.
The benefits of screening colonoscopy have largely been extrapolated from the results of RCTs of sigmoidoscopy, and from findings of observational studies that demonstrated a reduction in CRC incidence and mortality rates in participants who received colonoscopy23–25. The data from RCTs of sigmoidoscopybased screening, although differing in the number and frequency of assessments, endoscopic equipment used, and trial design, indicate that this approach is associated with reductions in CRC incidence rate by 18–23% and in diseasespecific mortality by 22–31%26. Of note, the reductions in the incidence rate and mortality were only statistically significant for distal cancers25, leading to the hypothesis that regular screening with colonoscopy would enable detection of as many distal cancers and more proximal cancers than screening with sigmoidoscopy, given the ability of colonoscopy to enable visualization of the colon proximal to the splenic flexure. Indeed, findings of two early multicentre trials on onetime colonoscopy screening for asymptomatic individuals indicated that sigmoidoscopy alone might result in a substantial burden of highrisk lesions being missed, as approximately 50% of these advancedstage neoplasms occurred in the proximal colon and were not associated with distal adenomas27,28. To date, no completed trial has directly compared the efficacy of sigmoidoscopy and colonoscopy, but pooled analyses of data from cohort studies on colonoscopy have revealed decreases in CRC incidence and mortality related to proximal and distal cancers25. These findings mirror the population decline in CRC incidence and mortality since the 1980s (FIG. 1); the sharpest decline in incidence rates occurs after 2000, when data from the above multicentre colonoscopy trials spurred increased uptake of colonoscopy screening. Colonoscopy every 10 years is considered by some experts to be the most-favourable screening strategy, given its sensitivity, ability to detect serrated polyps, and longlasting protection against future CRC29; however, other CRC screening strategies have also been shown to be effective, including sigmoidoscopy every 5 years and/or yearly stoolbased testing with faecal immunohistochemical or faecal occult blood tests30. Simulation models have estimated that the cumulative effect of the various CRC screening strategies is responsible for 50% of the observed decline in incidence and mortality rates of this disease in the USA31.
Screening for cervical cancer and CRC capitalizes on the typically slow, stereotyped progression that lesions comprising atypical cervical cells and colonic polyps undergo during their transformation into malignant neoplasms. The discovery of human papillomavirus (HPV) as the aetiological driver of most cervical cancers prompted further change in the approach to screening for this disease to incorporate consideration of HPV-infection status and adjust future interventions accordingly32. Cervical cells infected by oncogenic strains of HPV can sometimes develop into CIN, which can progress to cervical cancer if left untreated33. Similarly, some colonic polyps progress to malignancy after acquiring genetic mutations, which differ based on the histological type of the polyp; for example, investigators have demonstrated that hyperplastic polyps and tubulovillous polyps have distinct mutagenesis pathways34. The lead-time for such transitions spans several years, allowing adequate time for detection and treatment of the polyp before it becomes malignant. The findings regarding the biology of these diseases, and the experience in screening for them demonstrated that screening is most likely to be beneficial when the targeted cancer has a relatively uniform biology and a slower rate of progression (BOX 2: Lesson 1, corollary).
Another important lesson learned is that not all precancerous lesions are obligate precursors to invasive cancers; in fact, most are not (BOX 2: Lesson 2). Even in the absence of screening and removal, many cases of CIN do not progress to cervical cancer — the immune system often clears HPV infections associated with CIN grade 1, and 40% of CIN grade 2 lesions spontaneously regress32,35. Similarly, most colonic polyps will not transform into invasive neoplasms, and a substantial proportion — perhaps 30% — of small (<6–9 mm) polyps will regress, as suggested by findings of CTcolonography surveillance of unresected polyps36. Thus, many resected CIN lesions and colonic polyps would not have otherwise caused morbidity or death. Identification and removal of such lesions represents overdetection and overtreatment, respectively. Treatments for both of these lesion types are generally considered minimally invasive; nevertheless, they have inherent risks. Polypectomy to remove colonic polyps can rarely be complicated by bleeding or colonic perforation37, and colonoscopy can commonly lead to abdominal pain and bloating38. Excisional treatments for cervical lesions, such as loop excision and cone biopsy, carry risks, including bleeding and infection, and have been linked to adverse obstetrical outcomes, such as preterm birth39. Treatment harms are difficult to prove with certainty, and the increased risk of preterm birth among women who undergo the mostcommon cervical excisional technique (loop excision) has been called into question40. Nonetheless, current management guidelines recommend restraint in using excisional procedures for the treatment of cervical neoplasia in young women to avoid potential longterm health consequences associated with preterm birth.
Such risks, although not trivial, are generally tolerated because excisional treatments for CIN and colonic polyps are considered effective at preventing the development of invasive cancers, and are less toxic than the treatments that would otherwise be required if the diseases progressed to this stage (BOX 2: Lesson 2, corollary). Additionally, this practice is probably the predominant reason for the observed decline in the incidence of CRC and cervical cancers in the countries where screening is widespread (BOX 2: Lesson 3b). Tailoring the frequency of screening and limiting intervention for lesions that are not believed to be precursors to morbid disease, however, have been key challenges in screening aimed at prevention of these cancers. In guidelines published in 2012, the United States Preventive Services Task Force (USPSTF) recommend increasing the age of initiation of cervical screening cytology from 18 to 21 years, extending screening intervals, and implementing an upper age limit of 65 years for screening of women with prior negative test results32, reflective of a deeper understanding of the underlying biology of cervical neoplasia (BOX 2: Lesson 4).
On the other hand, the management of ductal carcinoma in situ (DCIS) of the breast has been the subject of heavy scrutiny precisely because current treatment strategies are not satisfying the corollary of Lesson 2 (BOX 2): treatment itself is associated with some risks, especially considering that the risk of progression and death for certain types of DCIS and invasive disease is quite low. The incidence of DCIS in the USA increased more than 500% between the early 1980s and late 1990s, largely paralleling the advent of screening mammography, and has stayed relatively constant since then41,42. That many cases of DCIS do not progress to invasive breast cancer is widely acknowledged; nevertheless, the standard therapy over the past 25 years or more has been surgical resection (mastectomy, or lumpectomy plus adjuvant radio-therapy) and hormonal therapy6,43. Despite treatment of >60,000 DCIS cases per year in the USA, the incidence of invasive breast cancer has not fallen42; moreover, breastcancer mortality has been unaffected by wide-spread treatment of DCIS (BOX 2: Lesson 3a)44. The natural history of DCIS is largely unknown, as most DCIS lesions are surgically resected. According to the available data, the prevalence of invasive cancer in the setting of DCIS might range from 0–50%45,46. Notably, the biology of the lesion dictates the risk of associated invasive cancer, with highgrade comedotype DCIS having a higher likelihood of coincident invasive cancer47.
Highgrade comedo and lowgrade noncomedo DCIS are increasingly recognized to represent distinct disease entities, with the latter probably constituting overdiagnosis. Lowgrade DCIS, even if untreated, is unlikely to cause breastcancerspecific mortality: a recent study reported 10year survival of 98.8% for women with untreated lowgrade DCIS, and 98.6% for those in whom lowgrade DCIS was surgically excised48. For lowgrade DCIS, the risk might be spread over the woman’s lifetime, whereas for high-grade DCIS, it might be concentrated within 5 years46. Indeed, highgrade DCIS is morecommonly associated with local recurrence after treatment, distant metastasis, and mortality, and could be considered a true precursor lesion49,50. Consideration of DCIS grade alone, however, is unlikely to be sufficient in determining the risk of invasive cancer, and could potentially continue to result in overdiagnosis. In the past 3 years, a geneexpression-profiling test has been introduced as a tool to delineate DCIS biology45. In addition, profiling of the tumour immune microenvironment might provide insights into the aetiology of, and inform treatment approaches for, the highest risk DCIS lesions51.
Early detection/stage shift
Screening approaches aimed at early detection of invasive cancer have been shown to reduce cancerrelated mortality rates in some large RCTs with longterm follow up; however, considerable controversy remains over optimal use of the screening tests, and regarding how to balance the benefits and the harms of overdiagnosis and subsequent overtreatment, especially in settings outside of closely monitored clinical trials. For example, mammographybased screening was shown to reduce breastcancerrelated mortality in early RCTs52–54, although morerecently available longterm followup data from completed trials have provided conflicting information on whether mammography decreases breastcancer mortality55,56. Of note, mammography trials have varied in key aspects, such as screening frequency and technique, randomization scheme, and attribution of outcome57. In metaanalyses of screening trials, investigators have reported a decrease in diseasespecific mortality associated with screening for breast cancer of approximately 20%, although the mortality reduction varies by age57,58: the absolute mortality reduction at 10 years is greatest in women aged 60–69 years (21 deaths per 10,000 women), and lowest in those aged 40–49 years (3 deaths per 10,000 women)59.
At the population level, breastcancer mortality in the USA has declined since 1990 (REF. 13). Despite some uncertainty, this decline is probably attributable to the combined effects of screening and therapy, and might be dominated by the unquestioned improvements in systemic therapy for locallyadvanced and node-positive breast tumours over the past two decades60. Microsimulations have yielded a very broad range of estimates for the contribution of screening to the decline in mortality observed in the USA (28–65%)61. The magnitudes of these estimates vary dramatically because simulations are influenced by the assumptions and inputs on which each model is based. In fact, even the lower bound estimate might be optimistic. As systemic treatments improve, the mortality reduction attributable to screening diminishes, and accurate modelling of the dissemination of new therapies, or the magnitude of their effects, can be difficult60. Likewise, accounting for overdiagnosis and lengthtime bias in models is challenging, leading to overestimation of the benefits of screening62. This consideration is important because 22–31% of breast cancers detected on mammography are estimated to represent overdiagnosis63.
Thus, two points relevant to screening can be made with the example of breast cancer. First, the mortality reduction attributable to screening diminishes as systemic treatments improve. Notably, most of the screening mammography trials were conducted before the advent of modern adjuvant treatment for breast cancer. Second, a reservoir of indolent disease exists that is detected with screening. After the widespread implementation of mammographic screening in the USA in the midtolate 1980s, the overall incidence of invasive breast cancer increased substantially, and remains substantially higher than rates before screening7 (FIG. 1). This increased incidence largely reflects detection of a greater number of localized (early stage) tumours, accompanied by a disproportionately small decrease in latestage cancers7, and whether this trend translates to lowering of diseaserelated mortality is controversial. Interestingly, an ecological study showed no reduction in breastcancerspecific mortality in regions of the USA with the highest uptake of mammographic screening64.
In the face of such complexity, the differing interpretation of the evidence by several guidelineissuing professional bodies around the world is perhaps unsurprising (TABLE 1). In updated guidelines published in February 2016, the USPSTF continued to recommend screening mammography every 2 years for women aged 50–74 years, and that women aged 40–49 years should only be offered screening based on individual circumstances related to patient preferences65. These recommendations were based, in part, on a decision analysis66 and systematic reviews59,67 commissioned by the USPSTF. In 2015, the American Cancer Society (ACS) modified their guidelines for breastcancer screening, based on a separate systematic review58, and their recommendations now more closely resemble the USPSTF guidelines, with the exception of recommended annual screening for women between the ages of 45 and 54 years68. American breastimaging societies and the American College of Obstetrics and Gynecology (ACOG) continue to recommend annual screening beginning at the age of 40 years69,70, whereas European countries recommend screening every 2–3 years, with starting ages that range between 40 and 50 years71–73.
Table 1 |.
Summary of mammography guidelines from selected nations
Country and organisation | Start screening at age (years) | Terminate screening at age (year) | Frequency of assessment | Comments |
---|---|---|---|---|
USA | ||||
United States Preventive Services Task Force (USPSTF)65 | 50 | 74 | Every 2 years (for women at average-risk of breast cancer) | Screening for women aged 40–49 years is a ‘grade C’ recommendation (‘offer or provide this service for selected patients depending on individual circumstances’) |
American Cancer Society (ACS)68 | 45 | As appropriate based on life expectancy | Annually then biennially at 55 years of age and older | Recommend continuing screening as long as the individual is in good health and has a life expectancy exceeding 10 years |
American College of Obstetricians and Gynecologists (ACOG)69 | 40 | As appropriate based on life expectancy | Annually | Suggest discussing cessation of screening with physician starting at age 75 |
American College of Radiology (ACR)/Society of Breast Imaging (SBI)70 | 40 | As appropriate based on life expectancy | Annually | Suggest continued screening as long as life expectancy exceeds 5–7 years |
Canada | ||||
Canadian Task Force on Preventive Health Care146 | 50 | 74 | Every 2–3 years | Not applicable |
Sweden | ||||
Socialstyrelsen73 | 40 | 74 | Every 18–24 months | Not applicable |
UK | ||||
National Health Service71 | 50 | 70 | Triennially | Expanding the age range of invited women to 47–73 years is being considered |
Netherlands | ||||
National Breast Screening Programme72 | 50 | 75 | Biennially | Not applicable |
Australia | ||||
Royal Australian College of General Practitioners147 | 50 | 74 | Biennially | Not applicable |
A similar picture is seen with screening for prostate cancer. Death from prostate cancer has also declined since the 1990s13, and this reduction is probably at least partially attributable to screening74. In the USA, the incidence of prostate cancer presenting initially as metastatic disease has decreased since the advent of PSAbased screening, indicating that screening and subsequent intervention does avert the progression of some localized tumours8. Nevertheless, two major RCTs of PSAbased screening produced discrepant findings related to prostatecancerspecific mortality: the Prostate, Lung, Colorectal, and Ovarian (PLCO)-study investigators reported no benefit75, whereas the European Randomized Study of Screening for Prostate Cancer (ERSPC) investigators reported a 21% reduction in the relative risk of prostatecancerspecific mortality76. Differences in the study designs and populations, as well as the relatively high proportions of men in the control groups who underwent PSAbased screening, might explain these conflicting results77. Regardless, the potential for overdiagnosis, with subsequent overtreatment, is widely recognized as a major downside of PSAbased screening. Indeed, a substantial increase in the incidence of prostate cancer has been observed following the dissemination of PSAbased screening (FIG. 1), mostly driven by early stage tumours with a low Gleason score7. Many lowgrade prostate cancers will not invade beyond the prostatic capsule during the man’s lifetime78, and thus subsequent biopsies, resections, and/or radiation therapy expose the patient to unnecessary harms. Additionally, a normal serum PSA level (typically below 4 ng/ml) does not exclude the possibility of prostate cancer: in the Prostate Cancer Prevention Trial79, 42.4% of all cancers with Gleason score ≥7 occurred in men with PSA values of ≤3 ng/ml. In the face of an unfavourable riskto-benefit ratio, the USPSTF has now recommended against the routine use of PSAbased screening, and to date, no country has introduced a national PSAbased screening programme80,81. Other major professional societies, however, urge shared decisionmaking regarding PSAbased screening. For example, the ACS recommends that this discussion should begin at the age of 50 years for men at average risk82, whereas the American Urological Association (AUA) recommends consideration of screening in men aged 55–69 years83. Similarly to the ACS, the European Association of Urology (EAU) recommends that PSA testing should be offered to men over 50 years of age (or earlier in certain risk groups, such as men with a family history of prostate cancer), and can continue until the individual’s life expectancy is less than 15 years84.
Lung cancer screening with LDCT has garnered increased attention based on results of the National Lung Screening Trial (NLST)85,86. In this study, 53,454 adults deemed to be at high risk of lung cancer on the basis of age and smoking history were randomly assigned to undergo three annual screenings with either LDCT or chest radiography85,86. After a median followup duration of 6.5 years, the LDCT arm had three fewer deaths per 1,000 individuals screened than the radiography arm — a 16% reduction in the relative risk of lungcancerspecific mortality86,87. An excess of 120 lung cancers was detected by LDCT versus radiography, however. With the use of modelling to account for life-time follow up, the overdiagnosis rate for screening with LDCT was estimated to be 11% overall, but was nearly 50% for bronchioloalveolarcell carcinoma and only 3% for other cell types88. The use of LDCT was also associated a cumulative falsepositive rate of 37% owing to the detection of benign pulmonary nodules that share imaging characteristics with lung cancer85. Results of a retrospective analysis of the NLST data, however, indicate that application of the LungRADS reporting system, developed by the American College of Radiology, could potentially reduce the falsepositive rate and overdiagnosis89. Findings of the Dutch–Belgian NELSON trial90 of screening for lung cancer with LDCT at 2year intervals after the initial screen indicated improved specificity compared with annual screening in the NLST85 (98.6% versus 73.4%), with the tradeoff of lower sensitivity (84.6% versus 93.8%). Nevertheless, a similar percentage of lung cancers were detected at stage 1 in the NELSON trial and the NLST85,90. Interval cancers comprised 35 out of 187 diagnosed lung cancers in the NELSON trial, although only 12 of these interval cancers (35%) were not visible on the prior screening scan90.
A concern is that the efficacy of LDCT seen in the clinicaltrial setting will not translate into effectiveness in community practice; some of the success in the NLST might be due to the high level of expertise in LDCT interpretation and patient management at the participating medical centres, 76% of which were National Cancer Institute (NCI)designated cancer centres91. Nevertheless, in the USA, screening for lung cancer is currently recommended for former or current smokers with a 30 packyear history of tobacco use (and a quit date within 15 years for former smokers) by the USPSTF and other professional societies92–94. Beginning screening at the age 55 years is generally advocated, but the recommended age at which to end screening varies between the guidelines94.
The careful delineation of the candidates for LDCT-based screening illustrates an understanding that not all individuals benefit equally from screening (BOX 2: Lesson 4). The prevailing lesson learned from current experience in screening of lung, breast, and prostate cancers, however, is that these cancers are truly heterogeneous in terms of their biological phenotype (BOX 2: Lesson 1). If the corollary of this lesson is not heeded, screening will disproportionately detect slower growing cancers and has the potential to reveal a reservoir of moreindolent disease. Given the clear excess of early stage cancers detected with populationlevel screening for breast and prostate cancers, room for improvement of these programmes clearly exists (BOX 2: Lesson 3a). Screening can lead to overdiagnosis and overtreatment if the potential for the detection of indolent cancers is not recognized and treatment decisionmaking does not account for disease biology. Geneexpression profiling of breast tumours, for example, has revealed a wide array of phenotypic features associated with differences in aggressiveness, and has begun to highlight the important interaction between biological phenotype and approaches to treatment95–97.
Tempering hype: an eye on improvement
The perception and message surrounding screening for cancer has evolved to acknowledge the complex interplay of risks and benefits inherent to its practice. Hype around screening initially centred around the sound bite that ‘early detection saves lives’ — an intuitive, powerful message, attractive to practitioners and patients alike. Early campaigns promoting the use of screening tests, such as mammography and colonoscopy, prominently featured (and in some cases, inflated) the purported benefits, while neglecting the potential harms98. Widereaching population screening was initiated at a time when the linear model of cancer progression prevailed. Reports from cancer registries showed that patients with early stage cancers had goodtoexcellent outcomes, and those with advancedstage disease had much higher mortality rates. This observation led to the belief that detecting cancer at an early stage would uniformly reduce cancerrelated mortality; however, this framework did not account for the extensive biological complexity and heterogeneity in cancer, which we are increasingly recognizing, or the associated variability in disease progression. Thus, the nearly uniform enthusiasm for screening contributed to a lowvalue, or ‘more is better’, approach to screening99. Admittedly, conceptualizing the rewards from less screening is difficult, and the lay public, based on decades of public-health messaging, tend to overestimate the benefits and underestimate the harms of screening100. Findings suggest that the concept of overdiagnosis, a clear harm that can be incurred in healthy, asymptomatic people, is discussed relatively infrequently between patients and healthcare providers101.
A guiding principle of cancer prevention and screening is that making healthy people better off than they already are is difficult. Prasad et al.102 have argued that no clear evidence indicates that any of the current cancer screening protocols convincingly reduce all-cause mortality, except LDCTbased screening for lung cancer — and even then, raise the possibility that the reduction in allcause mortality in the NLST might be smaller than reported. The downstream harms of overdiagnosis and overtreatment probably dilute or even nullify diseaserelated benefits of cancer screening in general, and exposure to such harms is more difficult to justify in the healthy population than in the management of patients with symptomatic disease. The frequency of screening should, therefore, be optimized based on detection of the tumour types for which beneficial outcomes of intervention are most likely. Those patients with tumours that progress too fast will not benefit from moreintensive screening, which would, however, increase the rates of falsepositive findings and overdiagnosis on the population level.
In Europe, such harms are ameliorated, to some extent, by the centralized approach to screening; programmes are organized with fixed budgets, and with formal consideration of the tradeoffs, as opposed to the opportunistic approach used in the USA. In each setting, the same data are viewed and interpreted through different metaphorical lenses — relating to, for example, the financing and organization of health care, malpractice litigation and cultural attitudes toward risk, interventions, and the politics behind the ‘war on cancer’. In Europe, such considerations have led to the generally moreconservative approach to the dissemination of screening. Consider breastcancer screening, for example: each European nation follows one guideline, and screening of women is usually recommended to begin at 50 years of age, occur every other year, and end at the age of 65–70 years (TABLE 1). Currently, no organized populationscreening programmes for lung or prostate cancer are active in Europe. Moreover, governmentbased screening in European nations affords several additional benefits. Firstly, comprehensive registries of screening outcomes are assembled. Secondly, quality measures can be better implemented, which probably explains the lower recall rates and higher cancertobiopsy ratios reported in Europe compared with the USA. Factors relevant to the latter advantage include the minimum requirement for mammogram reads (960 every 2 years in the USA compared with 5,000 per year in Denmark and the UK); double reading (having two radiologists review each image); and the centralization of reading, possibly making mammograms easier to compare, with an emphasis on high specificity103–105.
Nevertheless, important efforts are emerging in the USA to acknowledge the limitations and tackle the knowledge gaps with regard to cancer screening. These efforts have brought about renewed hope that screening programmes will meet the hype that initially accompanied them. First of all, increased awareness of overdiagnosis has prompted major professional groups to revise their guidelines68,106. Furthermore, the NCI convened a working group on overdiagnosis, which made several key recommendations to guide practice and research107. The American College of Physicians has also focused attention on highvalue care in cancer screening99,108. Moreover, increased coverage in the press and other laypublications in response to these actions has helped disseminate the screening debate among the general public.
Taking the key lessons learned from past experience and their corollaries (BOX 2), we can formulate corresponding action points to improve cancerscreening efforts. In the face of a heterogeneous disease biology (BOX 2: Lesson 1), efforts should be made to identify the true ‘targets’ of screening — namely, better defining a positive test result based on molecular phenotyping of lesions. Given the uncertainty regarding whether all precursor lesions are predecessors to clinically consequential disease (BOX 2: Lesson 2), a prevention or riskreduction strategy, rather than treatment intervention, should be considered as the initial approach for some of these lesions. Considering the heterogeneity of risk in the population (BOX 2: Lesson 3), risk stratification might better identify the individuals who are most likely to benefit from screening. Populationbased data on screening outcomes should be compiled into registries to provide continued feedback and thus enable quality improvement (BOX 2: Lesson 4). Lastly, similarly to treatment, screening should be based on both prognostic and predictive diagnostics, informed by a better understanding of disease phenotype, with a goal of characterizing and correlating screening abnormalities with the specific type of cancer biology using emerging prognostic and predictive tools. We posit that progress is being made across all five of these goals, with evidence of application and progress across all of the five cancers that are key targets for screening (that is, those of the breast, prostate, lung, cervix, and colon/rectum).
We have integrated the lessons learned with the screening ‘cascade’ proposed by Harris et al.109 to illustrate how tailored innovations are being incorporated at each step of the screening process (FIG. 2). We believe that such innovations set the stage for ‘precision screening’, which incorporates individualized riskprediction, based on clinical factors and biomarkers integrated with molecular characterization of the cancers detected. This approach should improve elucidation of the targets for cancer screening and prevention. Individualized data and patient values should be taken into account when making key decisions on whom to screen, when to initiate and cease screening, how often to screen, and what action to take for patients with abnormal findings. Efforts are already well underway to generate the information that will enable us to harness this knowledge to improve screening. The ‘output’ generated at each step of the screening cascade is linked with valuable opportunities for continued improvement. We have summarized the tools that will facilitate improvements in screening practices (BOX 3).
Figure 2. A framework for ongoing improvement of cancer-screening programmes.
We present a modified version of the screening cascade proposed by the High‑Value Care Task Force of American College of Physicians109. Our recommendations for cancer-screening programmes focus on incorporation of key clinical questions at each step of the cascade, as well as components of the ‘feedback loop’ (areas to refine) — aspects of screening decision‑making that can be actively improved using outcomes from the corresponding step on the cascade. IDLE, indolent lesions of epithelial origin.
Box 3 |. Toolkit for improving screening.
Site-specific tools
Risk-prediction models
Molecular-based tests to inform risk-stratification and treatment decisions
Feedback-based modification of screening interval and modality, and thresholds for initiating and stopping screening
Registry of outcomes as a resource for continued quality improvement
Standardization of test delivery and interpretation
Shared decision-making tools
Continued study of the biology, natural history, and treatment response of precancerous and cancerous lesions
Generalized strategies (applicable across all organ sites)
Integration of comorbidity assessment into decisions about screening, workup, and treatment
Common molecular classification of indolent tumours, for example, ‘IDLE’ (indolent lesions of epithelial origin) conditions — that is, redefinition of the term ‘cancer’
Screening systems that includes invitation to screen, recall, and outcomes tracking: ‘registry 2.0’
Precision along the screening cascade
Persons who are screened
Initiation of screening has to be undertaken acknowledging that “overdiagnosis exists and is common,” which is one of five recommendations made by an NCIsponsored thinktank working group on overdiagnosis107. The decision to screen should factor in an individual’s pretest probability of cancer, a threshold risk level at which testing is most likely to have a net benefit, and patient values and attitudes towards risk tolerance. Risk stratification has been practiced in a rudimentary form since the advent of screening, as the cumulative risk of nearly all cancers increases with age; therefore, minimum ages at which to begin screening in individuals at lowtoaverage risk have been recommended — be it faecal occult blood testing, sigmoidoscopy, or colonoscopy at the age of 50 years, or cervical cytology at 21 years of age. Differences in interpretation of the available evidence, however, continue to spur disagreement over these age thresholds110. Additionally, the presence of familial risk syndromes or a concurrent disease state associated with an elevated cancer risk places an individual in a highrisk group, warranting consideration of earlier and morefrequent screening. Examples include hereditary nonpolyposis colorectal syndromes111 or inflammatory bowel disease112,113 and CRC risk.
Beyond age and conditions associated with an increased risk of malignancy, exposure history is increasingly considered in riskstratification. For example, given the robust, dosedependent association between cigarette smoking and lung cancer, the NLST investigators selectively enrolled participants who met a minimum of 30 packyears of smoking history and, if former smokers, had quit less than 15 years before study entry85. Most screening guidelines and reimbursement criteria for lungcancer screening reflect the participant demographics of the NLST, namely limiting use of LDCT to people with a minimum smoking history of 30 pack-years92,114. Similar riskstratification tools have been developed for CRC screening115 and lungcancer screening116, and their clinical utility is currently being studied. Newer proposed algorithms for cervicalcancer screening suggest that HPV testing alone can identify a lowrisk population (those with a negative test result), or that typespecific testing of HPV types 16 and/or 18 might help to further refine riskstratification, such that women with evidence of oncogenic HPV types should have morediligent evaluation117,118.
Riskprediction models are increasingly being used for riskstratification. The Breast Cancer Risk Assessment Tool, one of the earliest risk prediction tools, was developed to identify women for inclusion in trials of preventive interventions for breast cancer, and considers exposure to endogenous hormones, in addition to other clinical risk factors119. Other riskprediction models are targeted at individuals suspected of having familial breast cancer120,121. The Breast Cancer Surveillance Consortium (BCSC) riskprediction tool incorporates age, race, family history, mammographic breast density, history of prior breast biopsy (and type of benign breast disease, if present) to calculate a woman’s 5year and 10year risks of developing breast cancer122,123. Beyond risk factors commonly incorporated in prediction models, some specific exposures clearly identify women at risk (for example, history of mantle radiation), and these women are recommended to undergo annual screening with MRI and mammography124. In addition, biomarkers have been combined with riskprediction tools in the hope of improving their performance. A polygenic risk score based on 76 single nucleotide polymorphisms (SNPs) has been shown to independently predict breast-cancer risk, and improved riskprediction when incorporated into the existing BCSC model125. To date, more than 90 SNPs have been associated with breastcancer risk126, and incorporation of additional SNPs might further enhance the predictive value of the polygenic risk score. In the upcoming WISDOM trial127, investigators will use the BCSC model, genetic mutation analysis, and a SNP panel to estimate the 5year breastcancer risk score of the women enrolled and, ultimately, assign them a tailored plan, personalizing the starting age, stopping age, and frequency of screening — all within the bounds of the USPSTF guidelines at study initiation. Over time, the risk model will be refined, as will screeningtest assignment127.
The screening test
Screening should follow another of the goals raised at the NCIsponsored thinktank: to “mitigate overdiagnosis by testing strategies that lower the chance of detecting unimportant lesions” (REF. 107). One can pursue this within three domains, as discussed in the following sections.
Choice of screening test.
Imaging tests serve to localize lesions and provide visual clues about the likelihood of malignancy and aggressiveness. With regard to prostatecancer screening, following up detection of an elevated PSA level with prostate MRI can help to rule out a falsepositive result, and if a lesion is present, to improve the yield of tumour tissue upon biopsy128. Women at very high risk of breast cancer, such as BRCAmutation carriers, firstdegree relatives of BRCAmutation carriers, or those with a 20–25% lifetime risk according to prediction models, should be screened annually with MRI, as an adjunct to mammography, given the superior sensitivity of MRI in this population124,129–131. Conversely, use of less invasive or costly strategies is a possibility for individuals on the other end of the risk spectrum. For example, less-frequent screening might be appropriate for individuals considered to be at ‘very low’ to ‘low’ risk of CRC according to the prediction model discussed in the previous section115. Of note, all current cervical screening guidelines by the ACS, USPSTF, and ACOG incorporate HPV testing as an alternative to cytologyonly strategies108.
Frequency of screening.
The frequency of testing is a question that has long been central to quality-improvement efforts in cervicalcancer and CRC screening. In both scenarios, results of the first test or previous tests are used to inform decisions about how and when to repeat screening. In cervicalcancer screening, a combination of a normal cytologytest result and a test result showing no evidence of infection with highrisk (oncogenic) HPV types among women aged ≥30 years predicts a particularly low risk of CIN and invasive cancer33; a 5year screening interval is currently recommended for these women32. Women with evidence of infection with oncogenic HPV types can have morediligent evaluation, whereas those with nononcogenic HPV infections can be followed less intensively117,118.
Likewise, the absence of colonic polyps on colonoscopy (and even the presence of small polyps that lack concerning histological features) is associated with a low risk of CRC development over the next decade, and the next screen can, therefore, occur in 10 years132. Breastdensity measurements obtained from initial mammograms (breast imagingreporting and data system (BIRADS) density) has been strongly linked to breastcancer risk; for example, extremely dense breasts in the setting of elevated risk, such as a family history of breast cancer, or in a woman aged 40–49 years support annual (rather than biennial) screening with mammography133. In the Stockholm3 (STHLM3 trial)134, a base-line PSA threshold of 1 ng/ml informed the frequency of prostate cancer screening: if a participant had a PSA level <1 ng/ml, he was not recommended to undergo screening during the following 6 years.
Definition of a positive test result.
Experience with precursor lesions has shown that not every ‘positive’ result warrants further immediate investigation or a biopsy. Historically, the standard of care for young women with abnormal cytology was followup colposcopy; however, newer screening approaches integrate watchful waiting (active surveillance). The 2012 management guidelines of the American Society for Colposcopy and Cervical Pathology135, for example, recommend that women aged 21–24 years with minimally abnormal cytological findings be followed with annual cytology testing, as many such lesions regress spontaneously. Incorporating strict criteria for embarking on a clinical workup into the screening cascade is important. Active surveillance, which will be covered in detail in the next section, can be used for individuals with indeterminate lesions, or those that probably represent indolent disease or its precursors. Additionally, according to the Lung-RADS reporting criteria, pulmonary nodules <6 mm in diameter detected on an initial LDCT screen do not constitute a ‘positive’ result given they do not require intervention, or necessitate changes to the screening frequency or modality136. This example illustrates an important concept, and one that is applicable to any screening study: a ‘finding’ does not necessarily constitute a ‘positive’ result. Lastly, results from the STHLM3 trial134 indicate that combining information on PSA levels, SNP genotype, circulating protein markers, and clinical variables can improve the accuracy of detection for prostate cancers with a Gleason score of ≥7. This demonstration that the STHLM3 model outperformed PSA testing alone for detection of these highrisk prostate cancers might usher in an era in which screening tests have morenarrowlydefined targets related to clinically consequential cancers134.
The clinical workup
The aim of a clinical workup in an individual with a positive screeningtest result is to establish a pathological diagnosis of cancer or highrisk neoplasia, and gather the data necessary for precision treatment. In many cases, a positive result will trigger an invasive diagnostic test, for example, an imageguided biopsy for a suspicious breast mass detected on mammography. In addition to standard pathological review for histology, extent of disease, and tumour markers, increasing options are available for molecular characterization of tumours. Geneexpressionprofiling tests have been developed to enable prediction of recurrence risk after treatment for invasive cancers and to support treatment decisions. Notable examples are the Oncotype DX® and MammaPrint® assays for geneexpression profiling of breast cancers95,96. Further refinements to these tests, such as establishment of an ‘indolent threshold’ for the MammaPrint® 70gene signature137, have enabled identification of a particularly indolent form of the disease. These advances have enabled geneexpression profiling to be performed on biopsy samples of screen-detected tumours to facilitate riskstratification and thus prevent overtreatment.
Molecular profiling has changed the view that a standard treatment is uniformly beneficial for all invasive cancers. Approximately onethird of breast cancers detected using modern screening modalities are defined as ‘ultralow risk’ based on geneexpression profiling138. These cancers are associated with no risk of breast-cancerrelated death in the first 15 years after surgical treatment and a <5% risk of late breastcancerrelated death (17–20 years after surgical treatment) with a short course of tamoxifen137. Certainly, identification of a precursor of this kind of indolent cancer has no rationale. Lowhistologicalgrade DCIS, as defined by pathologists, is probably a risk factor for the development of such indolent cancers, and this disease entity closely matches the definition of indolent lesions of epithelial origin, or ‘IDLE’ conditions, that was proposed by the working group convened by the NCI107. Other candidate IDLE conditions include the subset of indolent lung cancers identified within the NLST and Gleason 3 + 3 prostate cancers107. Setting up observational registries for IDLE conditions will enrich our understanding of the natural history of these tumours and provide guidance on how to incorporate information on disease dynamics (that is, whether the tumours progress, remain stable, or regress) into individualized management approaches. These efforts would parallel the NCI working group’s recommendation of creating observational registries for IDLE conditions107.
This approach has already been shown to hold promise with regard to lungcancer screening. The rollout of LDCT occurred in an era when the risks associated with screening and subsequent diagnostic testing were recognized, and as such, quality measures were formulated to standardize the clinical workup. For example, the LungRADS tool can be used to guide the management of nodules detected on LDCT136: on the basis of size, appearance, and growth rate, nodules are assigned a probability of malignancy using this tool, as well as a recommended timeframe and modality for surveillance. This strategy limits unnecessary imaging (only nodules larger than 6 mm, or 4 mm if new, require followup assessment) and tissue sampling, which is reserved for ‘Category 4B and 4X’ lesions, such as >1.5 cm solid nodules136. Likewise, the investigators of the NELSON study used strictly defined criteria for a ‘positive’ test result based on nodule volume or volumedoubling time, which probably improved the positive predictive value of LDCT (40.4%, 95% CI 35.9–44.7%), compared with the performance of this modality reported in other studies, such as the NLST (3.8%, 95% CI 3.4–4.3%)85,90.
The observation that CIN grade 2 lesions have a high spontaneous regression rate has led to recommendations that these lesions be followed, rather than treated, especially in young women in whom treatments might lead to adverse reproductive outcomes135. Repeating colposcopy with cytology at 6month intervals is specifically recommended for women aged 21–24 years, but can be offered to women of any age with CIN grade 2 in whom the harms of treatment are believed to outweigh the benefits135.
Treatment
A comprehensive discussion of cancer therapy is outside the scope of this Review, but tailored therapy is discussed briefly, in the context of limiting overtreatment of indolent tumours. Geneexpression profiling has deepened our understanding of the range of disease entities that are currently classified as ‘cancer’ based on the classic criteria of histological appearance. In the cases with diagnostic test results that suggest indolent disease, lessaggressive therapies should be pursued. For instance, lowgrade DCIS is more likely to be an indicator for an increased risk of future invasive cancer, similarly to its closely related pathological entity atypical ductal hyperplasia46, rather than an indication for immediate surgery and radiation therapy; a potentially better alternative is to consider these lesions as an opportunity for prevention, using selective oestrogenreceptor modulators or aromatase inhibitors. Thus, for certain women with breast lesions, endocrine therapy alone might be sufficient96,137.
Moreover, if the workup reveals an IDLE tumour, consideration should be given to active surveillance. When appropriate, changing the nomenclature of IDLE lesions, to reflect their typically benign clinical course, will help frame the decision between patients and providers. The NCIsponsored thinktank members recommended removing terms related to ‘cancer’ — as has been instituted for some CIN grade 3 lesions, formerly known as carcinoma in situ 107. A consortium of seven centres (funded by grants from the NCI) are working together to identify common biological criteria for indolent cancers and IDLE conditions, to help redefine ‘cancer’ in the era of modern molecular medicine139.
Systems-level improvements
Across the entire screening cascade, several advancements have the potential to improve screening programmes. For example, outcomes registries can support continued improvement by providing realtime feedback. National cancer registries have long been a main-stay in Europe, and have provided an opportunity for detailed cohort studies on screening outcomes140,141. In the USA, morelimited registries, such as the Breast Cancer Screening Consortium142, have linked data from regional mammography registries to form a representative sample of the country. The American College of Radiology’s lungcancerscreening registry represents a burgeoning attempt to form a national screening registry with an aim towards quality improvement143. The ultimate goals of this and similar ventures are to promote evidencebased practices (such as management of incidental findings) and improve reporting in order to enable continued assessment of screening practices. Participation in this registry enables screening centres to meet the qualityreporting requirements mandated by the Centers for Medicare and Medicaid Services143.
One key knowledge gap is centred on screening in the elderly and particularly those with considerable comorbidities — demographics in which few clinical trials of screening interventions have been conducted. Screening should proceed cautiously in the elderly, frail population; for example, many smokers aged within the 55–74 year range who represent the target for LDCT screening for lung cancer, based on the NLST results85,87, have concurrent cardiac or pulmonary disease that will limit their lifespan. Across the screening cascade, ideally, the individual’s underlying comorbidities and frailty should be incorporated into decisionmaking on the risk–benefit tradeoff. One such example of this approach is provided by ePrognosis, a prediction tool that is available online (http://eprognosis.ucsf.edu/) and as a smartphone application, and can be used to guide cancer screening in the elderly. The tool juxtaposes the predicted mortality benefit from screening with competing risks, based on a synthesis of published geriatric risk indices144. Integration of such tools into screening decisions is a promising area of future research, and the development of a tool that could be applied widely across all screening indications should be a research priority.
Finally, engaging individuals through shared decision-making and the routine offer of participation in studies should be major goals. Many decision tools have been created to facilitate discussions around screening, and tackle the complex interplay between risks, benefits, and each individual’s preferences145. Patientoriented studies, such as the WISDOM trial127, are probing the feasibility and acceptability of precision screening, and should provide critically needed data and key insights. Moreover, the Centers for Medicare and Medicaid Services has mandated that a “lung cancer screening counselling and shared decision making visit” must occur before a LDCT scan being ordered, and is a requirement for reimbursement, emphasizing the need to consider patient preferences143.
Patient preference could have an important role at the points in the screening cascade at which a biopsy or treatment is recommended. If potentially morbid disease is unlikely to be present, or the suspected lesion is thought to be associated with low mortality, then such uncertainties should be communicated to the patient. Patients’ values and levels of risk tolerance can help direct decision-making: those intolerant of the risks of a potential malignancy might favour an aggressive approach and, therefore, intervention, whereas others might favour a watchful-waiting approach. Those in the latter group should be cautioned of the potential need for morefrequent diagnostic testing, and the associated risks and benefits.
Conclusions
We now recognize that cancer encompasses a heterogeneous collection of conditions, and approaches to screening are changing accordingly. Opportunities for improvement are demonstrated by advancements in each of the screening programmes for lung, breast, prostate, colorectal, and cervical cancers, and can inform efforts to further advance the state of the art of screening. Learning who is at risk of which cancers, in terms of both site and biology, will be a critical underpinning for improvements in screening. The tools required to conduct studies to elucidate these data are coming online, owing to our increasing understanding of the genetic and biological basis of cancer risk, as well as the immunotypes, genotypes, and phenotypes of the tumours that arise. Herein, we have assembled the lessons learned from screening for five major cancers (breast, lung, prostate, cervical, and colorectal cancers; BOX 2) into a quality framework to accelerate our ability to introduce precision screening (FIG. 2), tailored to biology, patient preference, and clinical performance status.
Key points.
Tumours within any organ site can have a spectrum of biological phenotypes, ranging from indolent to highly aggressive
Screening for cancer is most likely to be beneficial when the target tumour type has a relatively uniform biology and a slower rate of progression
Not all precursor lesions are on an obligate pathway towards invasive-cancer development
Strategies for early detection of cancer must balance the benefits of mortality reduction (and reduction in invasive-disease incidence with screening for precancers) with the heterogeneity of the target disease and the consequent risk of overdiagnosis
Screening can be viewed as a ‘cascade’ involving multiple steps, such as selection of individuals to be screened, administration of the screening test, workup of positive findings, and, ultimately, treatment
Efforts are underway to individualize decision-making surrounding risk stratification, the modality and frequency of screening, and diagnostic and therapeutic interventions tailored to the biology of the detected tumour
Acknowledgements
We wish to thank Michael Pignone, MD, of the University of North Carolina at Chapel Hill, USA, for his thoughtful editing of this manuscript. We also wish to thank Alexandra Barratt, PhD, of the University of Sydney, Australia, for her helpful discussions and communications regarding cervical-cancer screening. Lastly, we wish to thank Mamta Shah of the University of California, San Francisco, USA, for her assistance in creating the draft figures for our manuscript.
Footnotes
Competing interests statement
M.E. is named on four patents applications for prostate-cancer diagnostics. G.F.S. is Principal Investigator of an NCI-funded grant that aims to identify the range of reasonable options for cervical-cancer screening from a patient-centred and economic perspective (R011CA169093). Y.S., W.C.B., B.S.K., and L.J.E. declare no competing interests.
References
- 1.Wilson JM & Jungner YG Principles and practice of screening for disease. Public Health Pap 34, 1–163 (1968). [Google Scholar]
- 2.Vogelstein B et al. Genetic alterations during colorectal-tumor development. N. Engl. J. Med 319, 525–532 (1988). [DOI] [PubMed] [Google Scholar]
- 3.Gyorffy B et al. Multigene prognostic tests in breast cancer: past, present, future. Breast Cancer Res 17, 11 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.National Cancer Institute. U.S. population data — 1969–2013 [online], http://www.seer.cancer.gov/popdata (2015).
- 5.Bleyer A & Welch HG Effect of three decades of screening mammography on breast-cancer incidence. N. Engl. J. Med 367, 1998–2005 (2012). [DOI] [PubMed] [Google Scholar]
- 6.Worni M et al. Trends in treatment patterns and outcomes for ductal carcinoma in situ. J. Natl Cancer Inst 107, djv263 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Esserman L, Shieh Y & Thompson I Rethinking screening for breast cancer and prostate cancer. JAMA 302, 1685–1692 (2009). [DOI] [PubMed] [Google Scholar]
- 8.Welch HG, Gorski DH & Albertsen PC Trends in metastatic breast and prostate cancer — lessons in cancer dynamics. N. Engl. J. Med 373, 1685–1687 (2015). [DOI] [PubMed] [Google Scholar]
- 9.Kirsh VA et al. Tumor characteristics associated with mammographic detection of breast cancer in the Ontario breast screening program. J. Natl Cancer Inst 103, 942–950 (2011). [DOI] [PubMed] [Google Scholar]
- 10.Shen Y et al. Role of detection method in predicting breast cancer survival: analysis of randomized screening trials. J. Natl Cancer Inst 97, 1195–1203 (2005). [DOI] [PubMed] [Google Scholar]
- 11.Woods WG et al. Screening of infants and mortality due to neuroblastoma. N. Engl. J. Med 346, 1041–1046 (2002). [DOI] [PubMed] [Google Scholar]
- 12.Ahn HS, Kim HJ & Welch HG Korea’s thyroid-cancer ‘epidemic’ — screening and overdiagnosis. N. Engl. J. Med 371, 1765–1767 (2014). [DOI] [PubMed] [Google Scholar]
- 13.Siegel RL, Miller KD & Jemal A Cancer statistics, 2016. CA Cancer J. Clin 66, 7–30 (2016). [DOI] [PubMed] [Google Scholar]
- 14.O’Grady TJ, Gates MA & Boscoe FP Thyroid cancer incidence attributable to overdiagnosis in the United States 1981–2011. Int. J. Cancer 137, 2664–2673 (2015). [DOI] [PubMed] [Google Scholar]
- 15.Peirson L, Fitzpatrick-Lewis D, Ciliska D & Warren R Screening for cervical cancer: a systematic review and meta-analysis. Syst. Rev 2, 35 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Whitlock EP et al. Liquid-based cytology and human papillomavirus testing to screen for cervical cancer: a systematic review for the U.S. Preventive Services Task Force. Ann. Intern. Med 155, 687–697 (2011). [DOI] [PubMed] [Google Scholar]
- 17.Sankaranarayanan R et al. Effect of visual screening on cervical cancer incidence and mortality in Tamil Nadu, India: a cluster-randomised trial. Lancet 370, 398–406 (2007). [DOI] [PubMed] [Google Scholar]
- 18.Sankaranarayanan R et al. HPV screening for cervical cancer in rural India. N. Engl. J. Med 360, 1385–1394 (2009). [DOI] [PubMed] [Google Scholar]
- 19.Shastri SS et al. Effect of VIA screening by primary health workers: randomized controlled study in Mumbai, India. J. Natl Cancer Inst 106, dju009 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Andrae B et al. Screening-preventable cervical cancer risks: evidence from a nationwide audit in Sweden. J. Natl Cancer Inst 100, 622–629 (2008). [DOI] [PubMed] [Google Scholar]
- 21.National Institute of Health. Cervical cancer. NIH Consensus Statement. NIH Consensus Development Program [online], https://consensus.nih.gov/1996/1996cervicalcancer102html.htm (1996).
- 22.Rositch AF, Nowak RG & Gravitt PE Increased age and race-specific incidence of cervical cancer after correction for hysterectomy prevalence in the United States from 2000 to 2009. Cancer 120, 2032–2038 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Nishihara R et al. Long-term colorectal-cancer incidence and mortality after lower endoscopy. N. Engl. J. Med 369, 1095–1105 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Brenner H, Chang-Claude J, Seiler CM, Rickert A & Hoffmeister M Protection from colorectal cancer after colonoscopy: a population-based, case–control study. Ann. Intern. Med 154, 22–30 (2011). [DOI] [PubMed] [Google Scholar]
- 25.Brenner H, Stock C & Hoffmeister M Effect of screening sigmoidoscopy and screening colonoscopy on colorectal cancer incidence and mortality: systematic review and meta-analysis of randomised controlled trials and observational studies. BMJ 348, g2467 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Weinberg DS & Schoen RE Screening for colorectal cancer. Ann. Intern. Med 160, ITC5–1 (2014). [DOI] [PubMed] [Google Scholar]
- 27.Lieberman DA et al. Use of colonoscopy to screen asymptomatic adults for colorectal cancer. Veterans Affairs Cooperative Study Group 380. N. Engl. J. Med 343, 162–168 (2000). [DOI] [PubMed] [Google Scholar]
- 28.Imperiale TF et al. Risk of advanced proximal neoplasms in asymptomatic adults according to the distal colorectal findings. N. Engl. J. Med 343, 169–174 (2000). [DOI] [PubMed] [Google Scholar]
- 29.Rex DK Colonoscopy: the current king of the hill in the USA. Dig. Dis. Sci 60, 639–646 (2015). [DOI] [PubMed] [Google Scholar]
- 30.United States Preventive Services Task Force. Screening for colorectal cancer: U.S. Preventive Services Task Force recommendation statement. Ann. Intern. Med 149, 627–637 (2008). [DOI] [PubMed] [Google Scholar]
- 31.Zauber AG The impact of screening on colorectal cancer mortality and incidence: has it really made a difference? Dig. Dis. Sci 60, 681–691 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Moyer VA Screening for cervical cancer: U.S. Preventive Services Task Force recommendation statement. Ann. Intern. Med 156, 880–891 (2012). [DOI] [PubMed] [Google Scholar]
- 33.Vesco KK et al. Screening for Cervical Cancer: A Systematic Evidence Review for the U.S. Preventive Services Task Force (Agency for Healthcare Research and Quality, 2011). [PubMed] [Google Scholar]
- 34.Grady WM & Markowitz SD The molecular pathogenesis of colorectal cancer and its potential application to colorectal cancer screening. Dig. Dis. Sci 60, 762–772 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.ASCUS-LSIL Triage Study (ALTS) Group. Results of a randomized trial on the management of cytology interpretations of atypical squamous cells of undetermined significance. Am. J. Obstet. Gynecol 188, 1383–1392 (2003). [DOI] [PubMed] [Google Scholar]
- 36.Pickhardt PJ et al. Assessment of volumetric growth rates of small colorectal polyps with CT colonography: a longitudinal study of natural history. Lancet Oncol 14, 711–720 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Levin TR et al. Complications of colonoscopy in an integrated health care delivery system. Ann. Intern. Med 145, 880–886 (2006). [DOI] [PubMed] [Google Scholar]
- 38.Ko CW et al. Incidence of minor complications and time lost from normal activities after screening or surveillance colonoscopy. Gastrointest. Endosc 65, 648–656 (2007). [DOI] [PubMed] [Google Scholar]
- 39.Kyrgiou M et al. Fertility and early pregnancy outcomes after conservative treatment for cervical intraepithelial neoplasia. Cochrane Database Syst. Rev 9, CD008478 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Conner SN et al. Loop electrosurgical excision procedure and risk of preterm birth: a systematic review and meta-analysis. Obstet. Gynecol 123, 752–761 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Ernster VL, Barclay J, Kerlikowske K, Grady D & Henderson C Incidence of and treatment for ductal carcinoma in situ of the breast. JAMA 275, 913–918 (1996). [PubMed] [Google Scholar]
- 42.Ozanne E et al. Characterizing the impact of 25 years of DCIS treatment. Breast Cancer Res. Treat 129, 165–173 (2011). [DOI] [PubMed] [Google Scholar]
- 43.Baxter NN, Virnig BA, Durham SB & Tuttle TM Trends in the treatment of ductal carcinoma in situ of the breast. J. Natl Cancer Inst 96, 443–448 (2004). [DOI] [PubMed] [Google Scholar]
- 44.Narod SA, Iqbal J, Giannakeas V, Sopik V & Sun P Breast cancer mortality after a diagnosis of ductal carcinoma in situ. JAMA Oncol 1, 888–896 (2015). [DOI] [PubMed] [Google Scholar]
- 45.Solin LJ et al. A multigene expression assay to predict local recurrence risk for ductal carcinoma in situ of the breast. J. Natl Cancer Inst 105, 701–710 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Sanders ME et al. Continued observation of the natural history of low-grade ductal carcinoma in situ reaffirms proclivity for local recurrence even after more than 30 years of follow-up. Modern Pathol 28, 662–669 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Leonard GD & Swain SM Ductal carcinoma in situ, complexities and challenges. J. Natl Cancer Inst 96, 901–920 (2004). [DOI] [PubMed] [Google Scholar]
- 48.Sagara Y et al. Survival benefit of breast surgery for low-grade ductal carcinoma in situ: a population-based cohort study. JAMA Surg 150, 739–745 (2015). [DOI] [PubMed] [Google Scholar]
- 49.Bijker N et al. Risk factors for recurrence and metastasis after breast-conserving therapy for ductal carcinoma-in-situ: analysis of European Organization for Research and Treatment of Cancer Trial 10853. J. Clin. Oncol 19, 2263–2271 (2001). [DOI] [PubMed] [Google Scholar]
- 50.Fisher ER et al. Pathologic findings from the National Surgical Adjuvant Breast Project (NSABP) eight-year update of Protocol B-17: intraductal carcinoma. Cancer 86, 429–438 (1999). [DOI] [PubMed] [Google Scholar]
- 51.Campbell MJ, et al. Characterizing the tumor immune microenvironment (TIME) in high-risk ductal carcinoma in situ [abstract]. Cancer Res 75, PD1–5 (2015). [Google Scholar]
- 52.Habbema JD, van Oortmarssen GJ, van Putten DJ, Lubbe JT & van der Maas PJ Age-specific reduction in breast cancer mortality by screening: an analysis of the results of the Health Insurance Plan of Greater New York study. J. Natl Cancer Inst 77, 317–320 (1986). [PubMed] [Google Scholar]
- 53.Nystrom L et al. Long-term effects of mammography screening: updated overview of the Swedish randomised trials. Lancet 359, 909–919 (2002). [DOI] [PubMed] [Google Scholar]
- 54.Tabar L et al. Efficacy of breast cancer screening by age. New results from the Swedish Two-County Trial. Cancer 75, 2507–2517 (1995). [DOI] [PubMed] [Google Scholar]
- 55.Miller AB et al. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial. BMJ 348, g366 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Moss SM et al. Effect of mammographic screening from age 40 years on breast cancer mortality in the UK Age trial at 17 years’ follow-up: a randomised controlled trial. Lancet Oncol 16, 1123–1132 (2015). [DOI] [PubMed] [Google Scholar]
- 57.Gotzsche PC & Jorgensen KJ Screening for breast cancer with mammography. Cochrane Database Syst. Rev 6, CD001877 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 58.Myers ER, Moorman P, Gierisch JM, Havrilesky LJ & Grimm LJ Benefits and harms of breast cancer screening: a systematic review. JAMA 314, 1615–1634 (2015). [DOI] [PubMed] [Google Scholar]
- 59.Nelson HD et al. Effectiveness of breast cancer screening: systematic review and meta-analysis to update the 2009 U.S. Preventive Services Task Force Recommend. Ann. Intern. Med 164, 244–255 (2016). [DOI] [PubMed] [Google Scholar]
- 60.Birnbaum J, Gadi VK & Markowitz E The effect of treatment advances on the mortality results of breast cancer screening trials: a microsimulation model. Ann. Intern. Med 164, 236–243 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 61.Berry DA et al. Effect of screening and adjuvant therapy on mortality from breast cancer. N. Engl. J. Med 353, 1784–1792 (2005). [DOI] [PubMed] [Google Scholar]
- 62.Autier P Efficient treatments reduce the cost-efficiency of breast cancer screening. Ann. Intern. Med 164, 297–308 (2016). [DOI] [PubMed] [Google Scholar]
- 63.Welch HG & Passow HJ Quantifying the benefits and harms of screening mammography. JAMA Intern. Med 174, 448–454 (2014). [DOI] [PubMed] [Google Scholar]
- 64.Harding C et al. Breast cancer screening, incidence, and mortality across US counties. JAMA Intern. Med 175, 1483–1489 (2015). [DOI] [PubMed] [Google Scholar]
- 65.Siu AL Screening for breast cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern. Med 164, 279–296 (2016). [DOI] [PubMed] [Google Scholar]
- 66.Mandelblatt JS, Stout NK & Schechter CB Collaborative modeling of the benefits and harms associated with different U.S. breast cancer screening strategies. Ann. Intern. Med 164, 215–225 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Nelson HD, Pappas M & Cantor A Harms of breast cancer screening: systematic review to update the 2009 U.S. Preventive Services Task Force recommendation. Ann. Intern. Med 164, 256–267 (2016). [DOI] [PubMed] [Google Scholar]
- 68.Oeffinger KC et al. Breast cancer screening for women at average risk: 2015 guideline update from the American Cancer Society. JAMA 314, 1599–1614 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 69.American College of Obstetricians–Gynecologists. Practice bulletin no. 122: breast cancer screening. Obstet. Gynecol 118, 372–382 (2011). [DOI] [PubMed] [Google Scholar]
- 70.Mainiero MB et al. ACR appropriateness criteria breast cancer screening. J. Am. Coll. Radiol 10, 11–14 (2013). [DOI] [PubMed] [Google Scholar]
- 71.Liston J & Wilson R (eds) Clinical guidelines for breast cancer screening assessment. GOV.UK [online], https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/465528/nhsbsp49June2010.pdf (2010).
- 72.de Jong N, Lock A, Carpay M & Hoebee B Dutch Breast Cancer Screening Program: organization and effectiveness [online], National Institute for Public Health and the Environment http://rivm.nl/dsresource?objectid=rivmp:239950&type=org (2014).
- 73.Socialstyrelsen. Screening för bröstcancer — rekommendation och bedömningsunderlag [online], http://www.socialstyrelsen.se/publikationer2014/2014-2-32 (2014).
- 74.Etzioni R et al. The prostate cancer conundrum revisited: treatment changes and prostate cancer mortality declines. Cancer 118, 5955–5963 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 75.Andriole GL et al. Mortality results from a randomized prostate-cancer screening trial. N. Engl. J. Med 360, 1310–1319 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 76.Schroder FH et al. Screening and prostate-cancer mortality in a randomized European study. N. Engl. J. Med 360, 1320–1328 (2009). [DOI] [PubMed] [Google Scholar]
- 77.Schroder FH & Roobol MJ ERSPC and PLCO prostate cancer screening studies: what are the differences? Eur. Urol 58, 46–52 (2010). [DOI] [PubMed] [Google Scholar]
- 78.Yin M, Bastacky S, Chandran U, Becich MJ & Dhir R Prevalence of incidental prostate cancer in the general population: a study of healthy organ donors. J. Urol 179, 892–895; discussion 895 (2008). [DOI] [PubMed] [Google Scholar]
- 79.Thompson IM et al. Prevalence of prostate cancer among men with a prostate-specific antigen level ≤4.0 ng per milliliter. N. Engl. J. Med 350, 2239–2246 (2004). [DOI] [PubMed] [Google Scholar]
- 80.Moyer VA Screening for prostate cancer: U. S. Preventive Services Task Force recommendation statement. Ann. Intern. Med 157, 120–134 (2012). [DOI] [PubMed] [Google Scholar]
- 81.Arnsrud Godtman R, Holmberg E, Lilja H, Stranne J & Hugosson J Opportunistic testing versus organized prostate-specific antigen screening: outcome after 18 years in the Goteborg randomized population-based prostate cancer screening trial. Eur. Urol 68, 354–360 (2015). [DOI] [PubMed] [Google Scholar]
- 82.Wolf A & Wender RC American Cancer Society guideline for the early detection of prostate cancer: update 2010. CA Cancer J. Clin 60, 70–98 (2015). [DOI] [PubMed] [Google Scholar]
- 83.Carter HB et al. Early detection of prostate cancer: AUA guideline. J. Urol 190, 419–426 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 84.Heidenreich A et al. EAU guidelines on prostate cancer. Part 1: screening, diagnosis, and local treatment with curative intent — update 2013. Eur. Urol 65, 124–137 (2014). [DOI] [PubMed] [Google Scholar]
- 85.National Lung Screening Trial Research Team. Results of initial low-dose computed tomographic screening for lung cancer. N. Engl. J. Med 368, 1980–1991 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 86.Pinsky PF, Church TR, Izmirlian G & Kramer BS The National Lung Screening Trial: results stratified by demographics, smoking history, and lung cancer histology. Cancer 119, 3976–3983 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 87.Aberle DR et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N. Engl. J. Med 365, 395–409 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 88.Patz EF Jr et al. Overdiagnosis in low-dose computed tomography screening for lung cancer. JAMA Intern. Med 174, 269–274 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 89.Pinsky PF et al. Performance of Lung-RADS in the National Lung Screening Trial: a retrospective assessment. Ann. Intern. Med 162, 485–491 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 90.Horeweg N et al. Detection of lung cancer through low-dose CT screening (NELSON): a prespecified analysis of screening test performance and interval cancers. Lancet Oncol 15, 1342–1350 (2014). [DOI] [PubMed] [Google Scholar]
- 91.Bach PB et al. Benefits and harms of CT screening for lung cancer: a systematic review. JAMA 307, 2418–2429 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 92.Moyer VA Screening for lung cancer: U.S. Preventive Services Task Force recommendation statement. Ann. Intern. Med 160, 330–338 (2014). [DOI] [PubMed] [Google Scholar]
- 93.Wender R et al. American Cancer Society lung cancer screening guidelines. CA Cancer J. Clin 63, 106–117 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 94.Tanoue LT, Tanner NT, Gould MK & Silvestri GA Lung cancer screening. Am. J. Respir. Crit. Care Med 191, 19–33 (2015). [DOI] [PubMed] [Google Scholar]
- 95.van ‘t Veer LJ et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 415, 530–536 (2002). [DOI] [PubMed] [Google Scholar]
- 96.Paik S et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N. Engl. J. Med 351, 2817–2826 (2004). [DOI] [PubMed] [Google Scholar]
- 97.Sorlie T et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc. Natl Acad. Sci. USA 98, 10869–10874 (2001). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 98.Woloshin S, Schwartz LM, Black WC & Kramer BS Cancer screening campaigns — getting past uninformative persuasion. N. Engl. J. Med 367, 1677–1679 (2012). [DOI] [PubMed] [Google Scholar]
- 99.Wilt TJ, Harris RP & Qaseem A Screening for cancer: advice for high-value care from the American College of Physicians. Ann. Intern. Med 162, 718–725 (2015). [DOI] [PubMed] [Google Scholar]
- 100.Hoffman TC & Del Mar C Patients’ expectations of the benefits and harms of treatments, screening, and tests: a systematic review. JAMA Intern. Med 175, 274–286 (2015). [DOI] [PubMed] [Google Scholar]
- 101.Moynihan R et al. Public opinions about overdiagnosis: a national community survey. PLoS ONE 10, e0125165 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 102.Prasad V, Lenzer J & Newman DH Why cancer screening has never been shown to ‘save lives’ — and what we can do about it. BMJ 352, h6080 (2016). [DOI] [PubMed] [Google Scholar]
- 103.Smith-Bindman R et al. Comparison of screening mammography in the United States and the United Kingdom. JAMA 290, 2129–2137 (2003). [DOI] [PubMed] [Google Scholar]
- 104.Kemp Jacobsen K et al. Comparing sensitivity and specificity of screening mammography in the United States and Denmark. Int. J. Cancer 137, 2198–2207 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 105.Esserman L et al. Improving the accuracy of mammography: volume and outcome relationships. J. Natl Cancer Inst 94, 369–375 (2002). [DOI] [PubMed] [Google Scholar]
- 106.US Preventive Services Task Force. Screening for breast cancer: U.S. Preventive Services Task Force recommendation statement. Ann. Intern. Med 151, 716–726 (2009). [DOI] [PubMed] [Google Scholar]
- 107.Esserman LJ et al. Addressing overdiagnosis and overtreatment in cancer: a prescription for change. Lancet Oncol 15, e234–e242 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 108.Sawaya GF et al. Cervical cancer screening in average-risk women: best practice advice from the Clinical Guidelines Committee of the American College of Physicians. Ann. Intern. Med 162, 851–859 (2015). [DOI] [PubMed] [Google Scholar]
- 109.Harris RP, Wilt TJ & Qaseem A A value framework for cancer screening: advice for high-value care from the American College of Physicians. Ann. Intern. Med 162, 712–717 (2015). [DOI] [PubMed] [Google Scholar]
- 110.Jorgensen KJ & Gotzsche PC Breast cancer: updated screening guidelines — much ado about small improvements. Nat. Rev. Clin. Oncol 13, 139–140 (2016). [DOI] [PubMed] [Google Scholar]
- 111.Stoffel EM et al. Hereditary colorectal cancer syndromes: American Society of Clinical Oncology Clinical Practice Guideline endorsement of the familial risk-colorectal cancer: European Society for Medical Oncology Clinical Practice Guidelines. J. Clin. Oncol 33, 209–217 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 112.Cairns SR et al. Guidelines for colorectal cancer screening and surveillance in moderate and high risk groups (update from 2002). Gut 59, 666–689 (2010). [DOI] [PubMed] [Google Scholar]
- 113.Kornbluth A & Sachar DB Ulcerative colitis practice guidelines in adults: American College of Gastroenterology, Practice Parameters Committee. Am. J. Gastroenterol 105, 501–523; quiz 524 (2010). [DOI] [PubMed] [Google Scholar]
- 114.Ruparel M & Navani N Fulfilling the dream. Toward reducing inequalities in lung cancer screening. Am. J. Respir. Crit. Care Med 192, 125–127 (2015). [DOI] [PubMed] [Google Scholar]
- 115.Imperiale TF, Monahan PO, Stump TE, Glowinski EA & Ransohoff DF Derivation and validation of a scoring system to stratify risk for advanced colorectal neoplasia in asymptomatic adults: a cross-sectional study. Ann. Intern. Med 163, 339–346 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 116.Tammemagi MC et al. Selection criteria for lung-cancer screening. N. Engl. J. Med 368, 728–736 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 117.Wright TC et al. Primary cervical cancer screening with human papillomavirus: end of study results from the ATHENA study using HPV as the first-line screening test. Gynecol. Oncol 136, 189–197 (2015). [DOI] [PubMed] [Google Scholar]
- 118.Huh WK et al. Use of primary high-risk human papillomavirus testing for cervical cancer screening: interim clinical guidance. Gynecol. Oncol 136, 178–182 (2015). [DOI] [PubMed] [Google Scholar]
- 119.Gail MH et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J. Natl Cancer Inst 81, 1879–1886 (1989). [DOI] [PubMed] [Google Scholar]
- 120.Tyrer J, Duffy SW & Cuzick J A breast cancer prediction model incorporating familial and personal risk factors. Stat. Med 23, 1111–1130 (2004). [DOI] [PubMed] [Google Scholar]
- 121.Mavaddat N, Rebbeck TR, Lakhani SR, Easton DF & Antoniou AC Incorporating tumour pathology information into breast cancer risk prediction algorithms. Breast Cancer Res 12, R28 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 122.Tice JA et al. Using clinical factors and mammographic breast density to estimate breast cancer risk: development and validation of a new predictive model. Ann. Intern. Med 148, 337–347 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 123.Tice JA et al. Breast density and benign breast disease: risk assessment to identify women at high risk of breast cancer. J. Clin. Oncol 33, 3137–3143 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 124.Saslow D et al. American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography. CA Cancer J. Clin 57, 75–89 (2007). [DOI] [PubMed] [Google Scholar]
- 125.Vachon CM et al. The contributions of breast density and common genetic variation to breast cancer risk. J. Natl Cancer Inst 107, dju397 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 126.Michailidou K et al. Genome-wide association analysis of more than 120,000 individuals identifies 15 new susceptibility loci for breast cancer. Nat. Genet 47, 373–380 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 127.Wisdom. About the study [online], https://wisdom.secure.force.com/portal/WsdSiteStudy (2015).
- 128.Schoots IG et al. Magnetic resonance imaging-targeted biopsy may enhance the diagnostic accuracy of significant prostate cancer detection compared to standard transrectal ultrasound-guided biopsy: a systematic review and meta-analysis. Eur. Urol 68, 438–450 (2015). [DOI] [PubMed] [Google Scholar]
- 129.Warner E et al. Surveillance of BRCA1 and BRCA2 mutation carriers with magnetic resonance imaging, ultrasound, mammography, and clinical breast examination. JAMA 292, 1317–1325 (2004). [DOI] [PubMed] [Google Scholar]
- 130.Kriege M et al. Efficacy of MRI and mammography for breast-cancer screening in women with a familial or genetic predisposition. N. Engl. J. Med 351, 427–437 (2004). [DOI] [PubMed] [Google Scholar]
- 131.Lee CH et al. Breast cancer screening with imaging: recommendations from the Society of Breast Imaging and the ACR on the use of mammography, breast MRI, breast ultrasound, and other technologies for the detection of clinically occult breast cancer. J. Am. Coll. Radiol 7, 18–27 (2010). [DOI] [PubMed] [Google Scholar]
- 132.Lieberman DA et al. Guidelines for colonoscopy surveillance after screening and polypectomy: a consensus update by the US Multi-Society Task Force on Colorectal Cancer. Gastroenterology 143, 844–857 (2012). [DOI] [PubMed] [Google Scholar]
- 133.Kerlikowske K et al. Outcomes of screening mammography by frequency, breast density, and postmenopausal hormone therapy. JAMA Intern. Med 173, 807–816 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 134.Grönberg H et al. Prostate cancer screening in men aged 50–69 years (STHLM3): a prospective population-based diagnostic study. Lancet Oncol 16, 1667–1676 (2015). [DOI] [PubMed] [Google Scholar]
- 135.Massad LS et al. 2012 updated consensus guidelines for the management of abnormal cervical cancer screening tests and cancer precursors. Obstet. Gynecol 121, 829–846 (2013). [DOI] [PubMed] [Google Scholar]
- 136.American College of Radiology. Lung CT Screening Reporting and Data System (Lung-RADS™) [online], http://www.acr.org/~/media/ACR/Documents/PDF/QualitySafety/Resources/LungRADS/AssessmentCategories.pdf (2015).
- 137.Lindstrom LS et al. MammaPrint accurately predicts long-term survival (25 years) and adjuvant tamoxifen therapy benefit in lymph node and negative patients (abstract). Cancer Res 75 (Suppl. 9), P4–11-12 (2015). [Google Scholar]
- 138.Esserman LJ et al. Impact of mammographic screening on the detection of good and poor prognosis breast cancers. Breast Cancer Res. Treat 130, 725–734 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 139.National Cancer Institute Division of Cancer Prevention. Consortium for Molecular Characterization of Screen-Detected Lesions Created: eight grants awarded [online], http://prevention.cancer.gov/news-and-events/news/consortium-molecular (2015).
- 140.Bennett RL, Blanks RG, Patnick J & Moss SM Results from the UK NHS Breast Screening Programme 2000–2005. J. Med. Screen 14, 200–204 (2007). [DOI] [PubMed] [Google Scholar]
- 141.Fracheboud J et al. Interval cancers in the Dutch breast cancer screening programme. Br. J. Cancer 81, 912–917 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 142.Breast Cancer Surveillance Consortium. Evaluating screening performance in practice NIH National Cancer Institute [online], http://breastscreening.cancer.gov/espp.pdf (2004).
- 143.Centers for Medicare & Medicaid Services. Decision memo for screening for lung cancer with low dose computed tomography (LDCT) (CAG-00439N) [online], https://www.cms.gov/medicare-coverage-database/details/nca-decision-memo.aspx?NCAId=274 (2015).
- 144.Yourman LC, Lee SJ, Schonberg MA, Widera EW & Smith AK Prognostic indices for older adults: a systematic review. JAMA 307, 182–192 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 145.Stacey D et al. Decision aids for people facing health treatment or screening decisions. Cochrane Database Syst. Rev 1, CD001431 (2014). [DOI] [PubMed] [Google Scholar]
- 146.The Canadian Task Force on Preventive Health Care. Recommendations on screening for breast cancer in average-risk women aged 40–74 years. CMAJ 183, 1991–2001 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 147.The Royal Australian College of General Practitioners. Guidelines for preventive activities in general practice, 8th edition. [online], http://www.racgp.org.au/your-practice/guidelines/redbook/