Challenges and Opportunities

in Environmental Epidemiology

of Cancer

Francine Laden, ScDAssociate Professor of Environmental Epidemiology

Harvard School of Public Health

and

The Channing Laboratory

Brigham and Women’s Hospital, Harvard Medical School

Structure of Talk

• Challenges of environmental

epidemiology

• Example: Diesel exhaust and lung

cancer

How do you define ENVIRONMENT?

Environmental Exposures are:

• Not genetic_______________________________________________________________________________________________________

• Pollution as opposed to life style

• Passive as opposed to active

• Involuntary as opposed to voluntary

• External as opposed to internal

A Few More Definitions:

• The ambient environment

– By the general population

• The occupational environment

– By the working population

Challenge:

Pollutants in the Ambient Environment are-

• Ubiquitous

– Hard to identify unexposed people

• Low levels

– Tight range

– Measurement error

• Passive

– Potentially unknown to participant

– Hard to identify unexposed AND exposed people

Challenge:

What is the Responsible Exposure?

“The Environment caused my cancer”

“Living in/working at _________

caused my cancer”

Challenge:

What to Measure:

• What is the chemical of interest?

• Is there a specific marker for your

exposure?

• What is the causative aspect of your

exposure?

• Is there a biomarker available?

Challenge:

When to Measure?

• Cancer is generally a disease of long

latency

• Timing during life cycle may be critical

• Usually performing measurements after

the fact

Challenge:

Who to Measure?

• Can you sample a representative

group?

• How do you define the boundaries of

your cluster (in both space and time)?

Example:

Diesel Exhaust Exposure and

Lung Cancer

Outline

• What is diesel exhaust?

• Overview of animal and human cancer studies

• The Trucking Industry Particle Study

Diesel Exhaust “a complex mixture of particles and gases”

• Vapor Phase

Carbon Monoxide

Carbon Dioxide

Sulfur Dioxide (from fuel)

Nitrogen Oxides

Aldehydes (odor)

Hydrocarbons

PAH Compounds

• Particulate Phase

Elemental Carbon

Sulfates

Hydrocarbons

PAH Compounds

PAH= Polycyclic Aromatic Hydrocarbons

Diesel Particle

Health Effects Institute, 1995

(Polycyclic Aromatic Hydrocarbons)

(PAH Compounds)

Elemental

Carbon (EC)

Organic

Carbon

(OC)

On surface

CO2 CO

SO2 Nitrogen oxides

What to Measure?

• Fine particle mass (PM2.5): < 2.5 µm

• PM1.0 mass: < 1.0 µm

• Elemental carbon core (EC mass)

• Organics on particle (OC mass)

• Gaseous pollutants (NO2 = vehicle exposure)

• PAHs (e.g. naphthalene)

• Particle number/surface area

Or Smell:

Animal Lung Cancer Studies

• Dose related increase in lung tumors at high levels of exposure (3500 µg/m3) in rats

• Negative results in other rodent species

• Results not specific to particles with associated organics (carbon particles )

• Mechanism: – In rats: overload of particle clearance mechanisms

and inflammatory changes precede the development of lung cancer

– But…inflammation is not part of the human lung cancer pathology

Human Lung Cancer Studies

• ~40 studies with 20%-50% elevated risk in diesel associated occupations– Truck Drivers

– Railroad Workers

– Bus Garage And Transport Workers

– Dock Workers

– Miners

• EPA, IARC, WHO

– likely or probable carcinogen

• California

– Toxic Air Contaminant (definite)

Previous Truck Driver Studies

solid circle = smoking adjusted open circle = smoking unadjusted

Decreased risk Increased risk“Null”

Statistically significant

• Diesel exhaust is likely to be carcinogenic to humans by inhalation and this hazard applies to environmental exposures

• Conclusions are based on the totality of evidence from human, animal, and other supporting studies

• Epidemiologic studies were done in occupational cohorts, but occupational and environmental levels overlap

How the Previous Epi Studies

Measure Exposure:

• Exposure definitions– Single job title or usual job

– Yearly job, years of employment

• Source of information– Self-report, census, next-of-kin, death certificate

– Union or work record

• Measurement of current exposure– Used to rank and validate exposure categories

– Not used in primary analysis

– No historical measurements

The Trucking Industry Particle Study

(TrIPS)

Funded by the National Cancer Institute

Research Questions

• Is lung cancer risk increased among

diesel exposed workers?

• What is the quantitative exposure-

risk relationship?

Population

• Four national unionized less-than-

truckload trucking companies

– Cooperation from both management and

labor (Teamsters)

Study Design

• Retrospective cohort study 1985-2000

• National assessment of current

exposures

• Smoking survey to representative

sample of workers

Epidemiologic Component

• From company records:

– Identified all unionized workers working ≥1 day in 1985 (58,326)

– Obtained personal work histories: chronological listing of job titles and work locations (terminal address, size)

• Mortality follow-up 1985-2000: NDI

– Date of death

– Cause of death – primary and underlying

Exposure Assessment Component

• Randomly selected 36 large terminals (>100 employees)

• Each was grouped with 1-2 smaller terminals located within 50 miles

• 7 day sampling trips

• Measured PM2.5, EC and OC in PM1.0 at each work location and upwind of the terminal

Trucking Terminals

Offices &

Lunchroom

Dock

Office

Forklift

Incoming

Trailers

Trailer

Tractor

(cab)

Freight

Outgoing

Trailers

Freight Terminal

Emission Sources

Yard

Repair

Shop

Dock

Wind

Pickup &

Delivery

Long Haul

(rural)

Air

Pollution

Emission sources

Pickup &

Delivery

Long Haul

(rural)

Tractors

(idling)

Offices &

Lunchroom

Dock

Office

Forklift

Freight

Yard

Repair

Shop

Dock

Dispatcher

Manager

Clerks

Dockman

Supervisor

Drivers:

Pickup &

delivery

Long-haul

Mechanic

Supervisor

Emission sources

WindAir

Pollution

Freight Terminal Operations

Jobs

Tractors

(idling)

Linking Exposure to Epidemiology

• Information available for everybody: job title,

terminal location and characteristics,

calendar year, years of work

• Develop statistical model to extrapolate

exposure for each individual

– Describes how above factors affect exposure

intensity

– Historic records used to estimate changes in

factors

Historical Exposures

• Traffic exposure in cities

– Trend for traffic emissions and traffic density

• Truck fleet changes with time

– Emissions by engine type, cab design, exhaust leaks

• Diesel fuel changes with time

– Sulfur, aromatic, and component contents

• Other exposure factors

– Changes in other sources

– Air pollution

Extrapolation of Historical Exposures

• Validation of the model

– Estimate 1989 exposures at sites of NIOSH surveys

– Data from older operations

• Potential limitations

– Absence of data on exposures before 1989

– Uncertainty in simulations of older exposures

– Missing company records

– Crude historic air pollution data

Comparisons to the

General Population

• Overall mortality:

– SMR=0.72 (95%CI=0.70-0.74)

• Lung cancer mortality

– Overall: SMR=1.04 (95%CI=0.97-1.12)

– All drivers: SMR=1.10 (95%CI=1.02-1.19)

– Dockworkers: SMR=1.10 (95%CI=0.94-

1.30)

Laden et al. EHP 2007

Smoking Rates:

Trucking versus General population

Cohort Study

• Exposure = job title

• Dose = cumulative years of work

• Comparison group = other job titles

Summary of Personal Exposure

Measurements (µg/m3)

EC PM2.5

N GM GSD GM GSD

Clerk 15 0.09 9.98 5.96 1.86

Dockworker 342 0.76 2.13 18.73 1.75

P&D 366 1.09 2.48 16.20 1.82

Long haul 173 1.12 1.91 19.96 2.30

GM=geometric mean; GSD=geometric standard deviation

Smith et al. JEM 2006

Lung Cancer HRs for +1 yrs of work

Job title PY Deaths HR (95%CI)*

Long haul 161,503 323 1.15 (0.92, 1.43)

P&D 139,054 233 1.19 (0.99, 1.42)

Dock 147,513 205 1.30 (1.07, 1.58)

Combo 96,543 150 1.40 (1.12, 1.73)

Mechanic 25,523 38 0.95 (0.66, 1.38)

Hostler 29,947 29 0.99 (0.68, 1.45)

Clerk 24,728 15 0.55 (0.32, 0.95)

Other jobs 13,040 12 0.89 (0.48, 1.63)

* Hazard ratio calculated using regression coefficients from a multivariate Cox proportional hazards

regression model stratified on age in 1985, decade of hire, and calendar time, with risk sets by attained

age, adjusted for the healthy worker survivor effect (total years on work, years off of work), race, and

census region.

Increased Risk with Years of Work

2.5 % Δ/yr wrk 3.6 % Δ/yr wrk

Long-haul drivers P&D drivers

Increased Risk with Years of Work

3.4 % Δ/yr wrk 4.0 % Δ/yr wrk

Dockworkers Combo Worker

Implications

• Increased risk of lung cancer for job

titles exposed to “freshly generated

exhaust”

• Easily extrapolated to the general

population driving on the same

roadways

Addressing the Challenges

• Ubiquitous, low level, passive exposures

• Responsible exposure

• What to measure

• When to measure

• Who to measure

Acknowledgments

• HSPH and HMS

• Eric Garshick

• Thomas Smith

• Mary Davis

• Ellen Eisen

• Jaime Hart

• Kevin Lane

• Serena Hon

• U of Wisconsin

• Jamie Shauer

• Advisory Board

• John Peters

• David Savitz

• Hans Kromhout

• Vicki Stover Hertzberg