Canadian National Breast Screening Study

From Wiki Journal Club
Jump to: navigation, search
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. 2014. 348:g366.
PubMed

Clinical Question

In women ages 40-59, does annual mammography improve breast cancer mortality compared to regular breast exams?

Bottom Line

Mammography had no mortality benefit compared to regular breast exams alone in women age 40-59 over 25 year follow-up, with higher rates of overdiagnosis.

Major Points

Guidelines on the subject of mammogram screening have variable recommendations on the frequency of screening for different age groups.

Mammogram screening was discovered to have mortality benefit in the Swedish Two-County Trial (which recruited patients starting in 1977), however this reduction in mortality was not replicated in the two largest randomized controlled trials to follow, the Canadian National Breast Screening Study (>80,000 women, age 40-59) and the Age Trial (>160,000 women age 40-49). Though the meta-analyses driving the guidelines found positive signal for mortality benefit, two large, high quality randomized controlled trials are negative and provide cause for skepticism and practice variation. Diagnosis of cancers that would otherwise not lead to death exposes patients to late and immediate risks of surgery, radiation, and chemotherapy which may negate the mortality benefit from early detection.

In the Canadian National Breast Screening Study, annual mammogram did detect more breast cancers than breast exam alone, however there was no concurrent improvement in mortality despite 25 years of follow-up data in a country with multiple national registry systems. This study raises significant concerns about a >20% overdiagnosis rate. While detected breast cancers in the mammogram group less likely to be large or node positive, this likely represents lead time bias given the lack of difference in breast cancer-related or overall mortality.

Further, retrospective analysis reveals that the first screening mammogram detected 73.3% of cancers diagnosed throughout the 5 year period, with significantly poorer survival rates than those diagnosed in the rest of the interval. Thus suggests that mammogram screening could potentially achieve a mortality benefit via initiation at an earlier age than 40, but results in overdiagnosis for women age 40-59. This trial does not provide data on women age 60-69.

Guidelines

USPSTF Breast Cancer Screening, January 2016
Recommends biennial screening for women age 50-74. Grade B.
The decision to start screening mammograms before age 50 should be individualized. Women age 40-49 may have a mortality benefit at the expense of false positive diagnoses and unnecessary biopsies. Grade C.
NCCN Breast Cancer Screening and Diagnosis, July 2016
Women age ≥ 40 who are asymptomatic and at average risk should have annual screening mammogram and clinical encounter. Class 1 recommendation.

Design

  • Multicenter randomized, controlled trial
  • N=89,835
    • Intensive (n=44,925)
    • Standard (n=44,910)
  • Setting: 15 centers in Canada
  • Enrollment: 1980-1985
  • Mean follow-up: 25 years
  • Analysis: Intention-to-treat
  • Primary outcome: Breast cancer mortality

Population

Inclusion Criteria

  • Women age 40-59

Exclusion Criteria

  • No history of breast cancer
  • No mammogram in past year
  • Not pregnant

Baseline Characteristics

Baseline characteristics not collected or described.

Interventions

  • Randomized to: annual mammogram + breast exam vs. breast exam alone

    • Clinical breast exam was performed annually for women age 50-59 and left to usual community care for women 40-49 by their family doctors.
    • Intervention was performed for 5 years
    • All patients thereafter returned to usual care
  • Outcomes were assessed at 5 years and at 25 years
    • Data collected via Canadian Cancer Registry, the Canadian national mortality database, direct patient surveys
    • In addition, every patient with abnormal exam or mammogram was referred to a study surgeon, who disclosed additional outcome data to the study annually

Outcomes

Experimental vs. control

Primary Outcomes

Breast cancer-related mortality
Overall: HR 0.99 (CI 0.88-1.12; P=0.87)
Age 40-49: HR 1.09 (CI 0.80-1.49; P=0.58)
Age 50-59: HR 1.02 (CI 0.77-1.36; P=0.88)
Overall mortality
HR 1.02 (CI 0.98-1.06; P=0.28)

Secondary Outcomes

Breast cancers diagnosed in 5 years
666 vs. 524 cancers (142 cases, or 27% overdiagnosis)
Size: 1.9 cm vs 2.1 cm (P=0.01)
30.6% vs 32.4% node positive (P=0.53)
Breast cancers diagnosed in 25 years
2584 vs 2609 cancers (106 extra cases, or 22% overdiagnosis)
70.6% vs. 62.8% for cancers diagnosed in mammogram vs. control group

Subgroup Analysis

Breast cancers diagnosed during 5 years of mammogram screening
73.3% of cancers detected at first screening
0.62 HR for survival for screen-detected cancer vs. interval-detected (P<0.01)
0.58 HR for survival of palpable cancer vs. image-only (P<0.0001)

Adverse Events

No additional adverse event data collected

Criticisms

  • While this trial was done in the era of adjuvant therapy, unlike prior trials, it fails to include more recent advances in breast cancer therapy which could improve outcomes from early detection
  • 14-15% loss to follow-up, while reasonable, is large compared to rare primary outcome and could result in significant bias if not random
  • Performing annual breast exam in women age 40-49 would help standardize results, improving internal validity
  • Defining specific screening strategies for the 20 years beyond the initial screening period would also standardize results, improving internal validity

Funding

Funding was provided by several named nonprofit and governmental sources, all of whom had no role in the preparation or submission of the article.

Further Reading

NCCN Guidelines: https://www.nccn.org/professionals/physician_gls/pdf/breast-screening.pdf