Key Programs
VA Oncology Conference - January 2002
January 24, 2002, Alexandria Virginia
Pamela M. Marcus, MS, PhD
Division of Cancer Prevention
National Cancer Institute
Lung cancer Screening: History
- 1950's and 60's
- Small uncontrolled studies
- No reduction in lung cancer mortality
- 1970's and 80's
- Three large randomized controlled trials (RCTs)
- No reduction in lung cancer mortality
Today's Talk
- Evidence underlying "not ready for prime time" view
- Data from Mayo Lung Project - over-diagnosis
- Data from Early Lung Cancer Action Project - not sufficient
- Lung Screening Study (LSS)
- National Lung Screening Trial (NLST)
Mayo Lung Project (MLP)
- RCT
- Began in early 1970s
- Nearly 10,000 male smokers
- Two study arms
- Intervention arm: chest x-ray and sputum cytology every 4 months for 6 years
- Usual care (control) arm: at trial entry only, a recommendation to receive same tests annually
MLP: Results, 1983
| Intervention | Usual Care |
Lung cancer cases | 206 | 160 |
Lung cancer deaths | 122 | 115 |
Lung cancer mortality rate (per 1,000 PY) | 3.2 | 3.0 |
|
MLP: Reactions
- Validity of finding questioned
- Limitation: follow-up time may have been too short to observe a mortality reduction
MLP: Additional Research
- Extended Lung Cancer Mortality Follow-up
- National Death Index Plus search
- Matches participant identifiers against state death certificates
- Provides coded cause of death
- Through 1996
MLP: Mortality - extended follow-up
| Intervention | Usual Care |
Lung cancer cases | 337 | 303 |
Lung cancer mortality rate (per 1,000 PY) * | 4.4 | 3.9 |
|
* p-value = 0.08
MLP: Case survival - extended follow-up
- Lung cancer cases:
- Intervention arm: 206
- Usual care arm: 160
- Cases diagnosed before the close of the project (July 1, 1983)
Back to Top
What is case survival?
Case Survival Vs. Mortality
- Case survival calculated using only the participants diagnosed with lung cancer
- Mortality calculated using all participants: those diagnosed with lung cancer and those who were not
MLP: Case Survival - Extended Follow-up
| Intervention | Usual Care |
Median survival time: all cancers | 1.3 years | 0.9 years |
Median survival time: early-stage resected cancers | 16.0 years | 5.0 years |
|
Case Survival: Interpretations
- Interpreted as some benefit of screening existed in MLP
- Incorrect interpretation
- In the absence of a mortality benefit, an improvement in case survival can only reflect screening biases
- Three possible screening biases: lead-time, length, over-diagnosis
MLP: Lead-time Bias?
MLP: Length Bias?
- Screening reveals a different set of tumors than are diagnosed as a result of symptoms
- "Length" refers to length of pre-clinical detection phase
- Unlikely explanation - total follow-up > 20 years on average
MLP: Over-diagnosis Bias?
- Most likely explanation
- Occurs when: Screening detects lesions that would never have been diagnosed during a person's lifetime in the absence of screening
- Two ways:
- Non-lung-cancer death prior to symptomatic detection (competing causes of mortality)
- Detection of indolent lesions
Over-diagnosis
- Value of screening is called into serious question
- Screening will cause harm that would not have occurred in the absence of screening
- Downstream effects may cause serious harm
Over-diagnosis: Is it plausible?
- Smokers also at high risk of other life-threatening events and diseases
- No consensus regarding whether a class of indolent lung cancer lesions exists
Back to Top
Does a class of indolent lung cancer lesions truly exist?
- High fatality rate of lung cancer - flawed argument
- Experience with other organ sites
MLP: Contributions
- An intense regimen of lung cancer screening may result in identification of lesions with little-to-no clinical relevance
- Particularly relevant information as we look towards spiral CT as a lung cancer screening modality
Spiral CT
- Used to diagnose lung cancer
- Low radiation dose spiral CT - detects abnormalities in asymptomatic high-risk persons
- ELCAP data (Lancet - July, 1999)
- 1,000 participants
- Each received low-dose CT and chest x-ray
ELCAP: Baseline Results
| Spiral CT | Chest X-ray |
Non-calcified nodules | 233 (23%) | 68 (7%) |
Malignant | 27 (3%) | 7 (1%) |
-----Stage I | 23 (85%) | 4 (60%) |
|
MLP and ELCAP
| MLP | ELCAP |
Greater detection? | Yes | Yes |
Earlier stage? | Yes | Yes |
Mortality reduction? | No | ? * |
|
*Not evaluable, due to absence of randomized control group
Randomized Controlled Trials (RCT)
- Most effective and least biased manner to assess screening modalities
- Random assignment of screening regimen
- Internal, yet separate, comparison group - provides the closest representation of what would have happened had screening not occurred
Lung Screening Study (LSS)
- Division of Cancer Prevention (NCI)
- Six PLCO screening centers - Washington DC, Minneapolis, Detroit, St. Louis, Birmingham, Marshfield (WI)
- PLCO: Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial
- Large RCT (n=154,958)
- Revisiting annual chest x-ray question
- LSS Participants:
- Males and females
- Ages 55-74
- Current and former (quit < 10 yrs) smokers
- At least 30 pack-years
- Not involved in PLCO
- Half randomized to CT; half to x-ray
LSS: Phase I
- Assessed feasibility of RCT
- September 2000 - August 2001
- One screen
- Overwhelming interest
- 3300+ participants
- Less than 2% ineligible due to recent CT
- Compliance greater than 94%
Back to Top
LSS: Additional Activities
- Began November 1, 2001
- Participants invited to receive second screen
- Same exam as first screening round
NCI Research - Status
- LSS - Inadequate power to detect small reduction in lung cancer mortality
- New project
- Approved by NCI's BSA (11/01)
- National Lung Screening Trial (NLST)
NLST Structure, January 2002
NLST - Trial Goals
- Primary: Determine whether lung cancer mortality is reduced by spiral CT compared to chest x-ray screening
- Secondary: Determine all-cause mortality, screening parameters
NLST - Design
- 46,000 participants randomized in 2 years
- 36,000 at PLCO sites
- 10,000 at ACRIN sites
- 50,000 in total (LSS included)
- Equal randomization
- Spiral CT versus chest x-ray
- Three annual screens
- Refer positives for follow-up
NLST - Statistical Power
| 2005 | 2006 | 2007 | 2008 | 2009 |
Mortality reduction | 52% | 37% | 28% | 24% | 21% |
Statistical power |
90% | 90% | 90% | 90% | 90% |
|
NLST - Eligibility Criteria
- 55 to 74 years old
- Smoking history
- 30 pack-years or more
- Current or former smoker (less than 15 years since quitting)
- Informed consent
NLST - Major Exclusion Criteria
- Previous lung cancer diagnosis
- Spiral CT scan of lungs or chest within past 18 mo.
- Current treatment for cancer other than non-melanoma skin
- Home oxygen supplementation
- Unexplained Sx: Wt loss > 15 lb. in 12 mo., hemoptysis
- Participation in another cancer screening or primary prevention trial (excluding smoking cessation)
NLST - Screening Results
- Sent to participant and physician within 3 weeks of exam
- Positive (suspicious for LC):
- Non-calcified nodule/mass - 4 mm
- Other abnormalities (or constellations of abnormalities) suggestive of lung cancer
- Positive screens: urged to seek medical care from a recognized specialist
Back to Top
Not Ready For Prime Time
- NLST will answer whether spiral CT can reduce the number of lung cancer deaths
- Until then - not ready for prime time
The Crux Of The Issue
If screening, in the absence of a benefit, were guaranteed to be innocuous, it would not be inappropriate to implement mass screening programs before definitive evidence supporting a mortality reduction was available. Unfortunately, that is not the case.
The "Costs" Of Screening
- Financial and otherwise
- Health care resources
- False positive tests
- Identification of lesions with little-to-no clinical relevance
The Unfortunate Truth
- Cancer screening will always result in harm, harm that would never have occurred in the absence of screening
- Benefit must outweigh harm
- Spiral CT must be properly evaluated
Collaborators
- Richard Fagerstrom
- Phil Prorok
- John Gohagan
- Members of the ACRIN team
Back to Top