At present, all of the evidence concerning the use of prophylactic mastectomy and oophorectomy for prevention of breast and ovarian cancer in high-risk women has been observational in nature; such studies are prone to a variety of biases, including case selection bias, family relationships between patients and controls, and inadequate information about hormone use. Thus, they may give an overestimate of the magnitude of benefit.
Screening is a means of detecting disease early in asymptomatic individuals, with the goal of decreasing morbidity and mortality. While screening can potentially reduce disease-specific deaths and has been shown to do so in cervical, colon, and breast cancer, it is also subject to a number of biases that can suggest a benefit when actually there is none. Biases can even mask net harm. Early detection does not in itself confer benefit. To be of value, screening must detect disease earlier, and treatment of earlier disease must yield a better outcome than treatment at the onset of symptoms. Cause-specific mortality, rather than survival after diagnosis, is the preferred endpoint (see below).
Because screening is done on asymptomatic, healthy persons, it should offer substantial likelihood of benefit that outweighs harm. Screening tests and their appropriate use should be carefully evaluated before their use is widely encouraged in screening programs, as a matter of public policy.
A large and increasing number of genetic mutations and nucleotide polymorphisms have been associated with an increased risk of cancer. Testing for these genetic mutations could in theory define a high-risk population. However, most of the identified mutations have very low penetrance and individually provide minimal predictive accuracy. The ability to predict the development of a particular cancer may some day present therapeutic options as well as ethical dilemmas. It may eventually allow for early intervention to prevent a cancer or limit its severity. People at high risk may be ideal candidates for chemoprevention and screening; however, efficacy of these interventions in the high-risk population should be investigated. Currently, persons at high risk for a particular cancer can engage in intensive screening. While this course is clinically reasonable, it is not known if it saves lives in these populations.
The Accuracy of Screening
A screening test's accuracy or ability to discriminate disease is described by four indices: sensitivity, specificity, positive predictive value, and negative predictive value (Table 82–2). Sensitivity, also called the true positive rate, is the proportion of persons with the disease who test positive in the screen (i.e., the ability of the test to detect disease when it is present). Specificity, or 1 minus the false positive rate, is the proportion of persons who do not have the disease and test negative in the screening test (i.e., the ability of a test to correctly identify that the disease is not present). The positive predictive value is the proportion of persons who test positive and actually have the disease. Similarly, negative predictive value is the proportion testing negative who do not have the disease. The sensitivity and specificity of a test are independent of the underlying prevalence (or risk) of the disease in the population screened, but the predictive values depend strongly on the prevalence of the disease.
Table 82–2 Assessment of the Value of a Diagnostic Testa |Favorite Table|Download (.pdf)
Table 82–2 Assessment of the Value of a Diagnostic Testa
|Condition Present||Condition Absent|
a = true positive
b = false positive
c = false negative
d = true negative
|Sensitivity||The proportion of persons with the condition who test positive: a/(a + c)|
|Specificity||The proportion of persons without the condition who test negative: d/(b + d)|
|Positive predictive value (PPV)||The proportion of persons with a positive test who have the condition: a/(a + b)|
|Negative predictive value||The proportion of persons with a negative test who do not have the condition: d/(c + d)|
Prevalence, sensitivity, and specificity determine PPV
Screening is most beneficial, efficient, and economical when the target disease is common in the population being screened. To be valuable, the screening test should have a high specificity; sensitivity need not be very high.
Potential Biases of Screening Tests
Common biases of screening are lead time, length-biased sampling, and selection. These biases can make a screening test seem beneficial when actually it is not (or even causes net harm). Whether beneficial or not, screening can create the false impression of an epidemic by increasing the number of cancers diagnosed. It can also produce a shift in proportion of patients diagnosed at an early stage and inflated survival statistics without reducing mortality (i.e., the number of deaths from a given cancer relative to the number of those at risk for the cancer). In such a case, the apparent duration of survival (measured from date of diagnosis) increases without lives being saved or life expectancy changed.
Lead-time bias occurs when a test does not influence the natural history of the disease; the patient is merely diagnosed at an earlier date. When lead-time bias occurs, survival appears increased, but life is not really prolonged. The screening test only prolongs the time the subject is aware of the disease and spends as a patient.
Length-biased sampling occurs because screening tests generally can more easily detect slow-growing, less aggressive cancers than fast-growing cancers. Cancers diagnosed due to the onset of symptoms between scheduled screenings are on average more aggressive, and treatment outcomes are not as favorable. An extreme form of length bias sampling is termed overdiagnosis, the detection of "pseudo disease." The reservoir of some undetected slow-growing tumors is large. Many of these tumors fulfill the histologic criteria of cancer but will never become clinically significant or cause death. This problem is compounded by the fact that the most common cancers appear most frequently at ages when competing causes of death are more frequent.
Selection bias must be considered in assessing the results of any screening effort. The population most likely to seek screening may differ from the general population to which the screening test might be applied. In general, volunteers for studies are more health conscious and likely to have a better prognosis or lower mortality rate, irrespective of the screening result. This is termed the healthy volunteer effect.
Potential Drawbacks of Screening
Risks associated with screening include harm caused by the screening intervention itself, harm due to the further investigation of persons with positive tests (both true and false positives), and harm from the treatment of persons with a true-positive result, even if life is extended by treatment. The diagnosis and treatment of cancers that would never have caused medical problems can lead to the harm of unnecessary treatment and give patients the anxiety of a cancer diagnosis. The psychosocial impact of cancer screening can also be substantial when applied to the entire population.
Assessment of Screening Tests
Good clinical trial design can offset some biases of screening and demonstrate the relative risks and benefits of a screening test. A randomized controlled screening trial with cause-specific mortality as the endpoint provides the strongest support for a screening intervention. Overall mortality should also be reported to detect an adverse effect of screening and treatment on other disease outcomes (e.g., cardiovascular disease). In a randomized trial, two like populations are randomly established. One is given the usual standard of care (which may be no screening at all) and the other receives the screening intervention being assessed. The two populations are compared over time. Efficacy for the population studied is established when the group receiving the screening test has a better cause-specific mortality rate than the control group. Studies showing a reduction in the incidence of advanced-stage disease, an improved survival, or a stage shift are weaker (and possibly misleading) evidence of benefit. These latter criteria are necessary but not sufficient to establish the value of a screening test.
Although a randomized, controlled screening trial provides the strongest evidence to support a screening test, it is not perfect. Unless the trial is population-based, it does not remove the question of generalizability to the target population. Screening trials generally involve thousands of persons and last for years. Less definitive study designs are therefore often used to estimate the effectiveness of screening practices. However, every non-randomized study design is subject to strong confounders. In descending order of strength, evidence may also be derived from the findings of internally controlled trials using intervention allocation methods other than randomization (e.g., allocation by birth date, date of clinic visit); the findings of cohort or case-control analytic observational studies; or the results of multiple time series studies with or without the intervention.
Screening for Specific Cancers
Widespread screening for cervical, colon, and breast cancer is beneficial for certain age groups. A number of organizations have considered whether or not to endorse routine use of certain screening tests. Because these groups have not used the same criteria to judge whether a screening test should be endorsed, they have arrived at different recommendations. The American Cancer Society (ACS) and the U.S. Preventive Services Task Force (USPSTF) publish screening guidelines (Table 82–3); the American College of Physicians (ACP) and the American Academy of Family Practitioners (AAFP) generally follow/endorse the USPSTF recommendations. Special surveillance of those at high risk for a specific cancer because of a family history or a genetic risk factor may be prudent, but few studies have assessed the influence on mortality.
Table 82–3 Screening Recommendations for Asymptomatic Normal-Risk Subjectsa |Favorite Table|Download (.pdf)
Table 82–3 Screening Recommendations for Asymptomatic Normal-Risk Subjectsa
|Test or Procedure||USPSTF||ACS|
Adults 50–75 years: every 5 years (“A”)b
Adults 76–85 years: “C”
Adults ≥85 years: “D”
|Adults ≥50 years: Screen every 5 years|
|Fecal occult blood testing (FOBT)|
Adults 50–75 years: Annually (“A”)
Adults 76–85 years: “C”
Adults ≥85 years: “D”
|Adults ≥50 years: Screen every year|
Adults 50–75 years: every 10 years (“A”)
Adults 76–85 years: “C”
Adults ≥85 years: “D”
|Adults ≥50 years: Screen every 10 years|
|Fecal DNA testing||“I”||Adults ≥50 years: Screen, but interval uncertain|
|Fecal immunochemical testing (FIT)||“I”||Adults ≥50 years: Screen every year|
|CT colonography||“I”||Adults ≥50 years: Screen every 5 years|
|Digital rectal examination (DRE)||No recommendation||Men ≥50 years, with a 10-year life expectancy; men ≥45 years, if African-American, or men with a first-degree relative diagnosed with prostate cancer <65 years; ≥40, if has several relatives with prostate cancer <65 years: Discuss and offer (with PSA testing) annually|
|Prostate-specific antigen (PSA)|
Men <75 years: “I”
Men ≥75 years: “D”
|As for DRE|
Women <65 years: Beginning 3 years after first intercourse or by age 21, screen at least every 3 years (“A”)
Women ≥65 years, with adequate, normal recent Pap screenings: “D”
Women after total hysterectomy for noncancerous causes: “D”
Women <30 years: Beginning 3 years after first intercourse or by age 21. Yearly for standard Pap; every 2 years with liquid test.
Women 30–70 years: Every 2–3 years if last 3 tests normal
Women ≥70 years: May stop screening if no abnormal Pap in past 10 years
Women after total hysterectomy for noncancerous causes: Do not screen
|Breast self-examination||“D”||Women ≥20 years: Breast self-exam is an option|
|Breast clinical examination||Women ≥40 years: “I” (as a stand-alone without mammography)|
Women 20–40 years: Perform every 3 years
Women ≥40 years: Perform annually
Women 40–49 years: The decision should be an individual one, and take patient context into account (“C”)
Women 50–74 years: every 2 years (“B”)
Women ≥75 years: (“I”)
|Women ≥40 years: Screen annually|
|Magnetic resonance imaging (MRI)||“I”|
Women >20% lifetime risk of breast cancer: Screen with MRI plus mammography annually
Women 15–20% lifetime risk of breast cancer: Discuss option of MRI plus mammography annually
Women <15% lifetime risk of breast cancer: Do not screen annually with MRI
|Complete skin examination||“I”||Self-examination monthly; clinical exam as part of routine cancer-related checkup|
Breast self-examination, clinical breast examination by a caregiver, mammography, and MRI have all been variably advocated as useful screening tools.
A number of trials have suggested that annual or biennial screening with mammography or mammography plus clinical breast examination in normal-risk women older than age 50 years decreases breast cancer mortality. Each trial has been criticized for design flaws. In most trials, breast cancer mortality rate is decreased by 15–30%. Experts disagree on whether average-risk women aged 40–49 years should receive regular screening (Table 82–3). The U.K. Age Trial, the only randomized trial of breast cancer screening to specifically evaluate the impact of mammography in women aged 40–49 years, found no statistically significant difference in breast cancer mortality for screened women versus controls after about 11 years of follow-up (RR, 0.83; 95% CI, 0.66–1.04); however, less than 70% of women received screening in the intervention arm, potentially diluting the observed effect. A meta-analysis of eight large randomized trials showed a 15% relative reduction in mortality (RR, 0.85; 95% CI, 0.75–0.96) from mammography screening for women aged 39–49 years after 11–20 years of follow-up. This is equivalent to a number needed to invite to screening of 1904 over 10 years to prevent one breast cancer death. At the same time, nearly half of women aged 40–49 years screened annually will have false-positive mammograms necessitating further evaluation, often including biopsy. Estimates of overdiagnosis range from 10 to 40% of diagnosed invasive cancers.
No study of breast self-examination has shown it to decrease mortality. A randomized controlled trial of approximately 266,000 women in China demonstrated no difference in mortality between a group that received intensive breast self-exam instruction and reinforcement/reminders and controls at 10 years of follow-up. However, more benign breast lesions were discovered and more breast biopsies were performed in the self-examination arm.
Genetic screening for BRCA1 and BRCA2 mutations and other markers of breast cancer risk has identified a group of women at high risk for breast cancer. Unfortunately, when to begin and the optimal frequency of screening have not been defined. Mammography is less sensitive at detecting breast cancers in women carrying BRCA1 and -2 mutations, possibly because such cancers occur in younger women, in whom mammography is known to be less sensitive. MRI screening may be more sensitive than mammography in women at high risk due to genetic predisposition or in women with very dense breast tissue, but specificity may be lower. An increase in overdiagnosis may accompany the higher sensitivity. The impact of MRI on breast cancer mortality with or without concomitant use of mammography has not been evaluated in a randomized controlled trial.
Screening with Papanicolaou smears decreases cervical cancer mortality. The cervical cancer mortality rate has fallen substantially since the widespread use of the Pap smear. Screening guidelines recommend regular Pap testing for all women who have reached the age of 21; some organizations advocate beginning earlier depending on sexual history. With the onset of sexual activity comes the risk of sexual transmission of HPV, the most common etiologic factor for cervical cancer. The recommended interval for Pap screening varies from 1 to 3 years. At age 30, women who have had three normal test results in a row may get screened every 2–3 years. An upper age limit at which screening ceases to be effective is not known, but women aged 65–70 years with no abnormal results in the previous 10 years may choose to stop screening. Screening should be discontinued in women who have undergone a hysterectomy for non-cancerous reasons.
Although the efficacy of the Papanicolaou smear in reducing cervical cancer mortality has never been directly confirmed in a randomized, controlled setting, a clustered randomized trial in India evaluated the impact of one-time cervical visual inspection and immediate colposcopy, biopsy, and/or cryotherapy (where indicated) versus counseling on cervical cancer deaths in women aged 30–59 years. After 7 years of follow-up, the age-standardized rate of death due to cervical cancer was 39.6 per 100,000 person-years in the intervention group versus 56.7 per 100,000 person-years in controls.
Fecal occult blood testing (FOBT), digital rectal examination (DRE), rigid and flexible sigmoidoscopy, colonoscopy, and CT colonography have been considered for colorectal cancer screening. Annual FOBT could reduce colorectal cancer mortality by a third. The sensitivity for fecal occult blood is increased if specimens are re-hydrated before testing, but at the cost of lower specificity. The false-positive rate for rehydrated FOBT is high; 1–5% of persons tested have a positive test. Only 2–10% of those with occult blood in the stool have cancer and 20–30% have adenomas. The high false-positive rate of FOBT dramatically increases the number of colonoscopies performed.
Fecal immunochemical tests appear to have higher sensitivity for colorectal cancer than nonrehydrated FOBT tests. Fecal DNA testing is an emerging testing modality; it appears to have increased sensitivity and comparable specificity to FOBT and could potentially reduce harms associated with follow-up of false-positive tests. The body of evidence on the operating characteristics and effectiveness of fecal DNA tests in reducing colorectal cancer mortality is limited.
Two case-control studies suggest that regular screening of those older than age 50 years with sigmoidoscopy decreases mortality. This type of study is prone to selection biases. A quarter to a third of polyps can be discovered with the rigid sigmoidoscope; half are found with a 35-cm flexible scope and two-thirds to three-quarters are found with a 60-cm scope. Diagnosis of adenomatous polyps by sigmoidoscopy should lead to evaluation of the entire colon with colonoscopy. The most efficient interval for screening sigmoidoscopy is unknown, but 5 years is often recommended. Case-control studies suggest that intervals of up to 15 years may confer benefit.
One-time colonoscopy detects ∼25% more advanced lesions (polyps >10 mm, villous adenomas, adenomatous polyps with high-grade dysplasia, invasive cancer) than one-time FOBT with sigmoidoscopy. Perforation rates are about 3/1000 for colonoscopy and 1/1000 for sigmoidoscopy. Debate continues on whether colonoscopy is too expensive and invasive for widespread use as a screening tool in standard-risk populations. Two observational studies suggest that efficacy of colonoscopy to decrease colorectal cancer mortality is restricted to the left side of the colon. CT colonography, if done at expert centers, appears to have a sensitivity for polyps ≥6 mm comparable to colonoscopy. However, the rate of extracolonic findings of abnormalities of uncertain significance that must nevertheless be worked up is high (∼15–30%); the long-term cumulative radiation risk of repeated colonography screenings is also a concern.
Chest x-ray and sputum cytology have been evaluated in randomized lung cancer screening trials. No reduction in lung cancer mortality has been seen, although all controlled trials have had low statistical power. Preliminary (unpublished) findings from the National Lung Screening Trial, a randomized controlled trial of screening for lung cancer in approximately 53,000 persons aged 55-74 years with a 30+ pack-year smoking history, have shown a statistically significant 20% reduction in lung cancer mortality in the spiral CT arm (354 deaths) compared to the chest x-ray arm (442 deaths). However, the mortality benefits must be weighed against the disadvantages of spiral CT for a given population. These include the potential radiation risks associated with multiple scans, the discovery of incidental findings of unclear significance, and a high rate of false-positive test results. Both incidental findings and false-positive tests can lead to invasive diagnostic procedures associated with anxiety, expense, and complications (e.g., pneumo- or hemothorax after lung biopsy).
Adnexal palpation, transvaginal ultrasound, and serum CA-125 assay have been considered for ovarian cancer screening. These tests alone and in combination do not have sufficiently high sensitivity or specificity to be recommended for routine screening of ovarian cancer. The risks and costs associated with the high number of false-positive results is an impediment to routine use of these modalities for screening. A large randomized controlled trial has shown that of female participants receiving at least one false-positive serum CA-125 test, 14% underwent a major surgical procedure (e.g., laparotomy with oophorectomy) for benign disease. For transvaginal ultrasound, the rate was close to 40%.
The most common prostate cancer screening modalities are DRE and serum prostate-specific antigen (PSA) assay. Newer serum tests, such as measurement of bound to free serum PSA, have yet to be fully evaluated. An emphasis on PSA screening has caused prostate cancer to become the most common non-skin cancer diagnosed in American males. This disease is prone to lead-time bias, length bias, and overdiagnosis, and substantial debate rages among experts as to whether it is effective. Prostate cancer screening clearly detects many asymptomatic cancers, but the ability to distinguish tumors that are lethal but still curable from those that pose little or no threat to health is limited. Men older than age 50 years have a high prevalence of indolent, clinically insignificant prostate cancers.
Two randomized controlled trials of the impact of PSA screening on prostate cancer mortality have been published. The Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial was a multicenter U.S. trial that randomized almost 77,000 men ages 55–74 years to receive annual PSA testing for 6 years or usual care. At 7 years of follow-up, no statistically significant difference in the number of prostate cancer deaths were noted between the arms (rate ratio, 1.13; 95% CI, 0.75–1.90). The data at 10 years (67% complete) showed similar results. Approximately 44% of men in the control arm received at least one PSA test during the trial, which may have potentially diluted an observed effect.
The European Randomized Study of Screening for Prostate Cancer (ERSPC) was a multinational study that randomized approximately 162,000 men between ages 50 and 74 years (with a predefined "core" screening group of men ages 55–69 years) to receive PSA testing every 4 years or no screening. Recruitment and randomizationprocedures and actual frequency of PSA testing varied by country. After a median follow-up of 9 years, a 20% relative reduction in the risk of prostate cancer death in the screened arm was noted in the "core" screening group (no difference in mortality was observed in the overall study population). The trial also found that 1140 men would need to be screened, and 48 additional cases treated to avert 1 death from prostate cancer.
The effectiveness of treatments for low-stage prostate cancer is under study. However, both surgery and radiation therapy may cause significant morbidity, such as impotence and urinary incontinence. Comparison of radical prostatectomy to "watchful waiting" in clinically diagnosed (not screen-detected) prostate cancers showed a small decrease in prostate cancer death rate in the surgery arm, but no statistically significant decrease in overall mortality was seen after 11 years of follow-up. Benefits were restricted to men younger than age 65 years. Urinary incontinence and sexual impotence were more common in the surgery arm. A man should have a life expectancy of at least 10 years to be eligible for screening. The USPSTF has found insufficient evidence to recommend prostate cancer screening for men younger than age 75 years; it recommends against screening for prostate cancer in men age 75 years or older ("D" recommendation) (Table 82–3).
Transvaginal ultrasound and endometrial sampling have been advocated as screening tests for endometrial cancer. Benefit from routine screening has not been shown. Transvaginal ultrasound and endometrial sampling are indicated for workup of vaginal bleeding in postmenopausal women but are not considered as screening tests in symptomatic women.
Visual examination of all skin surfaces by the patient or by a health care provider is used in screening for basal and squamous cell cancers and melanoma. No prospective randomized study has been performed to look for a mortality decrease. Unfortunately, screening is associated with a substantial rate of overdiagnosis.