1. CANCER SCREENING TESTS: EVALUATING THE EVIDENCE
Leah Karliner, MD, MAS
Department of Medicine
UCSF

2. CASE
56 y.o. man, healthy, no family history of GI cancer, no current symptoms of rectal bleeding, changes in stool or weight loss.
“Doc, can I get one of those virtual colonoscopies?”

3. OUTLINE Evaluating Tests Colon cancer screening: old tests and new
Breast cancer screening: mammograms and MRIs
Prostate cancer screening: should we screen?
Where to go for the evidence
(extra slides on ovarian and lung cancer screening)

4. PRINCIPLES OF SCREENING
Disease has high prevalence
Disease has serious consequences
Detectable preclinical phase
Treatment for presymptomatic disease is more effective than after symptoms develop
Positive impact on clinical health outcomes: early detection reduces cancer mortality

5. EFFECTIVENESS OF TEST
Tests should be simple, inexpensive and acceptable with a high sensitivity and specificity
Number of false positives is acceptably low

6. EFFECTIVENESS OF TEST
Questions to be answered when evaluating/comparing tests:
Who will be tested?
What tests will it supplement or replace?
Is the new test safer?
Is the new test less costly?
Is the test more specific (excluding cases of non-disease)?
Is the new test more sensitive (detecting more cases of disease)?
Is wide-spread use of the test feasible in practice?

7. SCREENING: OTHER CONSIDERATIONS
Screening in high risk groups
Selective vs universal screening
Rare diseases and false positive test results
Involving patients in the decision
What are the co-morbid conditions?
Associated life expectancy, feasibility of treatment, effects of treatment on quality of life

8. COLON CANCER

9. COLORECTAL CANCER: Principles of Screening
Disease has high prevalence: Second most common form of cancer in the U.S.
Disease has serious consequences: second highest cancer mortality rate overall in U.S.
Detectable preclinical phase – polyps
Treatment for pre-symptomatic disease is more effective than after symptoms develop - yes
Screening reduces cancer mortality:
Several studies have shown that screening with fecal occult blood test (FOBT) or sigmoidoscopy is associated with a reduction in colorectal cancer mortality
High prevalence disease with serious consequences.
Impact on health outcomes – reduces

10. HOW ARE WE DOING?
Adults > age 50, National Data from the Center for Disease Control:
FOBT in past 2 years
White
Black
Latino
Other
Multiracial
Ever had a sig or colonoscopy
27%
28%
24%
17%
20%
53%
54%
49%
39%
41%
BRFSS, 2004

11. COLON CANCER SCREENING RECOMMENDATIONS
U.S. Preventive Services Task Force recommends screening all persons over 50
Benefits of screening outweigh potential harms
Quality of evidence, magnitude of benefit and potential harms vary with each method
Unclear which is the best test: FOBT, FOBT plus sigmoidoscopy, colonoscopy

12. AVAILABLE TESTS
Tests should be simple, inexpensive and acceptable with a high sensitivity and specificity: ????
Available/commonly used tests:
Fecal occult blood test
Sigmoidoscopy
Colonoscopy
Newer tests:
Virtual Colonoscopy?
Fecal DNA testing?
Immunochemical FOBT?

13. WHICH TEST? Are the tests equally safe? Are the tests equally costly?
Are the tests equally specific?
Are the tests equally sensitive?
Is wide-spread use of the test feasible in practice?

14. TEST ISSUES Sigmoidoscopy FOBT Fair evidence for reducing mortality
Sigmoidoscopy alone can miss proximal neoplasia – a positive test needs to be followed by colonoscopy
FOBT
Good evidence for reducing mortality
Trials used 6 sample every 1-2 years
Positive test needs to be followed by colonoscopy
Many providers use digital FOBT as a primary screening test - this is different use from in the trials - is in office single stool sample testing enough?

15. FOBT vs. IN-OFFICE SINGLE FOBT
Sensitivity for advanced neoplasia was 24% for 6 sample FOBT vs 5% for digital FOBT
Specificity was 94% for 6 sample FOBT and 98% for digital FOBT
Digital FOBT is a poor screening method
Collins, 2005

16. IS COLONOSCOPY “BETTER?”
Two observational studies of patients undergoing colonoscopy
Goal: Determine prevalence and location of colonic neoplasia in asymptomatic patients and the risk of proximal advanced neoplasia in patients with or without distal neoplasia
Did NOT assess morbidity and mortality

17. IS COLONOSCOPY “BETTER?”
Colonoscopy showed some lesions that would have been missed by sigmoidoscopy alone
distal polyps were a predictor of proximal neoplasia,
but some patients with proximal neoplasia did not have distal polyps
If sigmoidoscopy alone had been done and if every adenomatous polyp triggered colonoscopy, 80% of high risk lesions would have been detected

18. SCREENING COLONOSCOPY?
Would proximal lesions have been detected by FOBT?
No assessment of morbidity and mortality

19. SCREENING COLONOSCOPY?
More sensitive than FOBT/sigmoidoscopy
More specific than FOBT
Higher risk (diagnostic colonoscopies have 1/2000 perforation rate; with polypectomy 1/ )
More costly? (USPSTF says all of these screening methods are probably cost-effective)
Presumed to save lives because used as diagnostic test in FOBT studies, but at higher rate than FOBT?
Feasibility in practice dependent on availability of gastroenterologists and insurance coverage

20. WHICH TEST?
Most preventable cases of colon cancer are found in those who have never been screened
If we screened with the currently available tests at the recommended intervals, we could make a big impact – particularly in ethnic minorities
Any screening is better than no screening for reducing colorectal cancer mortality

21. NEWER TESTS
Virtual Colonoscopy
Fecal DNA
Immunochemical FOBT (iFOBT)

22. VIRTUAL COLONOSCOPY
Non-invasive colon imaging method using thin section CT
Test characteristics in largest study to date
N=1,233 average risk individuals
Sensitivity
94% for polyps ≥8 mm
89% for polyps ≥6 mm
Specificity
96% for polyps ≥10 mm
80% for polyps ≥6 mm
Pickhardt, 2003

23. VIRTUAL COLONOSCOPY
Study used 3 D technology which is not available everywhere
Single center study
Are these results reproducible? Is this feasible in widespread practice?

24. VIRTUAL COLONOSCOPY Multicenter study of screening population
615 participants at 9 hospitals
Two-dimensional scans
Sensitivity
55% for lesions ≥10 mm
39% for lesions ≥6 mm
Specificity
96% for lesions ≥10 mm
91% for lesions ≥6 mm
Cotton, 2004

25. VIRTUAL COLONOSCOPY Requires bowel prep and insufflation
Patients do not necessarily prefer over colonoscopy
Test interpretation is very time consuming
Cost effectiveness
Assuming 100% sensitivity and specificity
To replace colonoscopy, it would have to be less than 50% the cost of colonoscopy and compliance would have to be 15-20% better
Sonnenberg, 1999

26. FECAL DNA TESTING
DNA alterations in colorectal cancer can be detected in the stool
Potential advantages
Non-invasive
No preparation
Detection along entire length of the colon

27. FECAL DNA TESTING
Recently evaluated as a screening test in asymptomatic individuals aged 50 and older
Fecal DNA test (21 mutations), Hemoccult II and colonoscopy
4404/5486 completed all three aspects of the study
Subgroup of 2507 patients were analyzed
Imperiale, 2004

28. FECAL DNA TESTING Fecal DNA Hemoccult II
Sensitivity for invasive cancer
51.6%
12.9%
Sensitivity for invasive cancer/adenoma with high grade dysplasia
40.8%
14.1%
Sensitivity for advanced neoplasia
18.2%
10.8%
Specificity

29. FECAL DNA TESTING
20% of the subjects either did not provide samples or did not have colonoscopy
Many were aged 65 and over
Both FOBT and fecal DNA had relatively low sensitivities compared with what was expected based on prior studies

30. FECAL DNA: REMAINING QUESTIONS
Are health outcomes improved?
Even if we assume benefit based on FOBT trials, how much?
Do the benefits outweigh the risks?
Public expectations about accuracy of DNA testing?
Frequency of testing?
Acceptability and availability?
Cost
$400 to $800 vs $3 to $40 for FOBT

31. IMMUNOCHEMICAL FOBT Potential advantages: Easier to use
Improved specificity
Probably small increase in sensitivity (may not need as many samples)
Test characteristics in large average risk populations has not been studied

32. COLORECTAL CANCER SCREENING: CONCLUSIONS
Any currently available screening is better than no screening for reducing colorectal cancer mortality
Virtual colonoscopy, immunochemical tests and fecal DNA testing may have a role in the future

33. Breast Cancer

34. BREAST CANCER SCREENING
Disease has high prevalence: most commonly detected cancer in women in U.S.
but lower prevelance for women in 40’s
Disease has serious consequences: second highest cancer mortality rate for women in U.S.
Detectable preclinical phase – microcalcifications
Treatment for pre-symptomatic disease is more effective than after symptoms develop – unclear in case of DCIS
Screening reduces cancer mortality: Several studies have shown that screening mammography can reduce mortality
RCTs have not shown a mortality reduction in women in their 40’s

35. USPSTF Data are most clear for women aged 50-69
United States Preventive Services Task Force recommends screening mammography with or without clinical breast examination every 1-2 years for women aged 40 and older
Data are most clear for women aged 50-69
For women in their forties the evidence is weaker
Benefit to women aged 70 and older if life expectancy not compromised by co-morbid disease

36. USPSTF
Evidence insufficient for or against clinical breast examination alone
Evidence insufficient for or against teaching or performing routine breast self-examination

37. TEST ISSUES
Tests should be simple, inexpensive and acceptable with a high sensitivity and specificity: Increased density of pre-menopausal breast tissue leads to decreased sensitivity
Number of false positives is acceptably low:
Cumulative risk of false positive result: 49% after ten mammograms
False positive rates were higher for women in their forties than for women age 50-69
(Elmore et al, 1998)

38. MAMMOGRAPHY IN WOMEN AGED 40-49
Increased density of premenopausal breast tissue leads to decreased sensitivity
More cases discovered by mammography in women in their forties are ductal carcinoma in situ (DCIS) than in women in their fifties (40-45% vs 20%)
Clinical significance of DCIS is unclear
Only 20% will progress to invasive cancer over 10 years and those that do progress will do so slowly
Who will benefit from DCIS treatment?

39. MAMMOGRAPHY IN WOMEN AGED 40-49
Discuss the pros and cons of mammography screening and should consider patient risk factors in making a decision about screening
If mammography is offered, it should be performed annually

40. MAMMOGRAPHY IN ELDERLY WOMEN
Age is the most important risk factor for breast cancer
Nearly half (47%) of breast cancer is diagnosed in women over the age of 65 and 52% of breast cancer mortality occurs in this age group
Competing mortality
Mandelblatt, JGIM 2005

41. SCREENING HIGH RISK WOMEN
Women with BRCA1 and BRCA2 mutations or with a family history of breast cancer are often diagnosed at a young age
Screening is often offered to younger high risk women but efficacy is not known
Lower sensitivity of mammography in younger women
High tumor growth rate
Atypical mammography changes in women with BRCA mutations

42. MRI SCREENING Does MRI have a role for screening in high risk women?
Sensitive method of breast imaging and has been used as a diagnostic tool in women with breast cancer
Not influenced by breast density
Specificity is variable
Expensive

43. MRI SCREENING
236 Canadian women aged with BRCA1 and BRCA2 mutations had 1-3 annual screening examinations
MRI, ultrasound, mammography annually
Clinical breast examination every 6 months
Warner et al JAMA 2004

44. MRI SCREENING 22 cancers detected
6 DCIS
All four screening modalities combined had a sensitivity of 95% vs 45% for mammography plus clinical breast exam

45. SENSITIVITY AND SPECIFICITY
Test
Sensitivity
Specificity
MRI
77%
95%
Mammography
36%
99.8%
Ultrasound
33%
96%
Clinical Breast Exam 9%
99%

46. IMPACT FOR CLINICAL PRACTICE
MRI may be useful in screening high risk women, although the effect of MRI screening on mortality is not yet known
MRI is not currently recommended for screening average risk women

47. Prostate Cancer

48. PROSTATE CANCER: SHOULD WE SCREEN?
Disease has high prevalence: Most commonly diagnosed cancer in U.S. men
10% lifetime risk
30% of men have prostate cancer at autopsy
Disease has serious consequences: variable; prostate cancer may be a benign disease for many men
Detectable preclinical phase – ?PSA
Treatment for pre-symptomatic disease is more effective than after symptoms develop - Does early detection do more good than harm or vice versa?
Complications of prostate cancer treatment
8.4% incontinence
60% impotence
Prostate Cancer Outcomes Study 24 month follow up
Screening reduces cancer mortality: ???

49. IS TREATMENT OF EARLY DISEASE EFFECTIVE?
Does treatment of early prostate cancer reduce morbidity and mortality?
RCT showed reduction in mortality, prostate cancer mortality, metastatic disease and local progression
Absolute reduction in mortality is small
Bill-Axelson, NEJM 2005

50. DOES SCREENING DECREASE MORTALITY? EPIDEMIOLOGIC EVIDENCE
Prostate cancer mortality has decreased following the introduction of prostate cancer screening
Reduction in mortality followed an initial increase in incidence
Due to PSA screening?
Short time interval
Changes in treatment
Is the decline most in the areas with most screening?

51. RANDOMIZED CLINICAL TRIALS
46,000 men underwent DRE and PSA
11 year follow-up
23 % of invited group and 6.5% of non-invited group underwent screening
Decrease in prostate cancer mortality, but small numbers of deaths overall
Labrie, Prostate 2004
PLCO trial sponsored by the NCI is ongoing
European Randomized Study of Screening for Prostate Cancer

52. DIGITAL RECTAL EXAMINATION
One-third of prostate cancers occur in areas which can be reached
Higher sensitivity performed by urologists
An abnormal digital rectal examination increases the likelihood of prostate cancer somewhat
A negative examination does not change the likelihood of a clinically significant prostate cancer
Low sensitivity of the examination

53. PSA SCREENING: TEST ISSUES
15% of men over the age of 50 have an elevated PSA
PSA >10 ng/ml:
66% have prostate cancer
PSA 4-10 ng/ml:
22% have prostate cancer

54. PSA SCREENING: TEST ISSUES
Levels of 4.0 ng/ml or less have typically been considered to be normal
Recent results from the Prostate Cancer Prevention Trial show that prostate cancer is not rare even in these men
27% cancer in those with PSA 3.1 to 4.0
24% in those with PSA 2.1 to 3.0
17% in those with PSA 1.1 to 2.0
10% in those with PSA 0.6 to 1.0
7% in those with PSA up to 0.5
How many cancers would be clinically important?

55. PSA SCREENING: MODIFICATIONS
PSA Density
PSA Velocity
Free and complexed PSA
So far, none of these modifications have proven superior to PSA alone

56. PROSTATE CANCER SCREENING: RECOMMENDATIONS
USPSTF: insufficient evidence to recommend for or against routine screening for prostate cancer using PSA or DRE
PSA can detect early prostate cancer, but inconclusive evidence about whether early detection improves health outcomes
ACP: individualize the decision to screen after discussion with patient about potential benefits and harms

57. SUMMARY OF RECOMMENDATIONS
Colon cancer: any screening is better than no screening, use commonly available tests
Breast cancer:
women aged 50 to 69 should undergo mammography every 1-2 years
discuss the pros and cons of mammography screening with women aged and over age 70
MRI screening may be useful in high risk women
Prostate cancer: discuss pros and cons of PSA with eligible men; await PLCO trial

58. SUMMARY OF RECOMMENDATIONS
Cervical cancer:
Begin screening w/in 3 years of onset of sexual activity or at age 21;
Screen at least every 3 years;
Stop screening at age 65 in low risk women with adequate screening history (USPSTF 2003)
‘reflex’ HPV testing on pap smears read as ASCUS (ACOG 2003)
Ovarian cancer:
maybe in high risk women only; await PLCO trial
women at high risk should consider oral contraceptives (37% reduction in incidence)
Lung cancer: do not screen; await PLCO trial
Smoking Cessation!

59. WHERE TO GO FOR THE EVIDENCE
U.S. Preventive Services Task Force
Technology Evaluation Center / Blue Cross - Blue Shield Association
California Technology Assessment Forum / Blue Shield of California Foundation

60. OVARIAN AND LUNG CANCER SCREENING

61. OVARIAN CANCER: SHOULD WE SCREEN?
Lifetime risk of ovarian cancer
No affected relatives 1.2%
One affected relative 5%
2 affected relatives 7%
Hereditary syndrome 40%
Ovarian cancer limited to the ovaries is associated with a much higher survival rate

62. OVARIAN CANCER: SCREENING TECHNIQUES
Serum CA-125 assay
Trans-vaginal ultrasound
Serum CA-125 plus ultrasound

63. OVARIAN CANCER SCREENING: CLINICAL TRIAL
22,000 U.K. women
Annual screening vs no screening for 3 years with 7 year follow-up
Screening
CA-125
Ultrasound if elevated CA-125
Surgical evaluation if ultrasound abnormal
Slight increase in mean survival
No difference in mortality
Jacobs et al, Lancet 1999

64. OVARIAN CANCER SCREENING: CONCLUSIONS
Many women must be screened to detect a few cases
Small increase in survival:
Is it worth it?
Low disease prevalence limits utility of the tests despite high sensitivity and specificity

65. SCREENING RECOMMENDATIONS
USPSTF: potential harms outweigh potential benefits
NIH Consensus Conference: Insufficient evidence
Many organizations recommend annual pelvic examination
No evidence
Although there are no data regarding screening in high risk women, experts recommend:
annual screening with rectovaginal pelvic examination, CA-125 and transvaginal ultrasound

66. FUTURE DIRECTIONS PLCO Trial
74,000 women aged 55-74
CA-125 at entry and annually for 5 years
Annual transvaginal ultrasound
13 year follow-up
United Kingdom Trial of Ovarian Cancer Screening
200,000 women with 7 year follow-up
Lysophosphatidic Acid (LPA)
Tumor marker which shows promise

67. LUNG CANCER: SHOULD WE SCREEN?
Lung cancer is the number one cause of cancer mortality in both men and women
If screening works for so many other cancers, why don’t we screen for lung cancer?

68. LUNG CANCER SCREENING: SYSTEMATIC REVIEW
Does screening for lung cancer reduce lung cancer mortality
Included 7 trials of lung cancer screening
Frequent screening with chest x-rays was associated with an increase in mortality
RR 1.11 (95% C.I )
No difference in chest X-ray plus cytology vs chest X-ray alone

69. ROLE OF CT No evidence that screening CT reduces mortality
Lung Cancer Screening Study
NCI sponsored
High risk patients
CT or chest X-ray
Results available soon
At this time, spiral CT should not be routinely used in clinical practice

70. USPSTF RECOMMENDATIONS
Evidence is insufficient
Screening with x-ray, low dose CT, sputum cytology or combination can detect lung cancer early, but there is no evidence that any screening strategy reduces lung cancer mortality.

71. PRIMARY PREVENTION OF LUNG CANCER
Smoking cessation
Smoking cessation!!!!!