Trends in Breast Cancer in Younger Women in
Contrast to Older Women
Benjamin F. Hankey, Barry Miller, Rochelle Curtis, Carol Kosary*
The SEER Program currently consists of 11 population-
Using data from the Surveillance, Epidemiology, and End
based registries under contract with the NCI to provide data on
Results (SEER) Program and the National Center for
all cancers diagnosed in residents of their coverage areas. Data
Health Statistics, trends in female breast cancer rates were
collected include cancer site/type, morphology, extent of dis-
examined for the time period 1973-1989 for the age group
ease. first course of cancer-directed therapy, and patient fol-
20-39 and contrasted with those for older ages. Only about
low-up, including cause of death. Data are submitted to the
7% of breast cancers occur by the age of 40; the risk of
NCI annually. The coverage areas of the 11 registries are the
developing breast cancer prior to the age of 40 is less than
entire states of Hawaii, New Mexico, Iowa. Utah, and Connec-
1%. The incidence trends for women in the 20-39 age group
ticut and the metropolitan areas of San Francisco, Seattle,
have been essentially stable. whereas for women 40 and
Detroit, and Atlanta. The county of Los Angeles and the San
older the rates increased steeply during the 1980s (at a faster
Jose-Monterey area of California, which is adjacent to the five-
rate than anticipated based on historical trends) and then
county San Francisco area, have been recently added to the
leveled off beginning in 1987. Breast cancer mortality has
SEER Program; however, data from these registries are not in-
been much more stable over time than incidence. Up to age
cluded in the analysis presented here.
40, blacks have a higher incidence than whites. Over age 40,
Mortality data were obtained from the National Center for
white rates exceed those for blacks, and the absolute and
Health Statistics for the entire United States and population
relative differences in incidence increase with advancing
estimates by county from the Census Bureau, the latter being re-
age. For whites, 5-year relative survival rates improved with
quired for the calculation of cancer incidence and mortality rates.
advancing age up to age 50. Blacks under the age of 30 had sur-
In calculating incidence/mortality rates for age ranges, e.g..
vival rates similar to whites, whereas, in the older age groups,
20-39, the rates are age-adjusted within the designated range
whites had somewhat better survival rates overall and by stage.
using weights from the 1970 United States Standard population
The occurrence of second cancers was also analyzed in women
for the 5-year age groups included in the range.
with a first invasive breast cancer. Cancers found to occur at
higher than expected rates included leukemia and cancers of
Results
the breast, ovary, and lung. [Monogr Natl Cancer Inst 16:7-14,
1994]
Proportion of Breast Cancers Diagnosed at Younger Ages
Fig. 1 presents the cumulative distribution of breast cancers
occurring in women during the period 1987-1989 by age at
Summary statistics on breast cancer incidence and mortality.
diagnosis in the SEER areas. Only 6.5% of the breast cancers
e.g., the age-adjusted rate. are heavily weighted by the higher
were diagnosed in women under age 40. and 21.8% were diag-
rates in older age groups. Since breast cancer rates in younger
nosed before the age of 50. Applying the 6.5% to the 183 000
women display patterns that are different from those in older
breast cancers expected to be diagnosed in women in the United
women. it is important to review rates by age group to discern
States during 1993 (2) indicates that approximately 12 000
patterns that are hidden by the summary measures.
breast cancers will be diagnosed in women under the age of 40.
Probability of Developing Breast Cancer
Data Sources
The probability of developing cancer can be calculated up to
a specified age or for an entire lifetime. It is a useful statistic to
Data on newly diagnosed cancer cases. including patient fol-
consider, particularly in regard to the study of a rare event like
low-up. were collected through the Surveillance. Epidemiol-
ogy, and End Results (SEER) Program. The SEER Program
began in 1973 as part of the National Cancer Program and is
administered by the Surveillance Program in the Division of
*Affiliations of authors: B.F. Hankey. B. Miller, C. Kosary. (Cancer Statistics
Cancer Prevention and Control at the National Cancer Institute
Branch, Surveillance Program, Division of Cancer Prevention and Control). R.
(NCI). The mission of the SEER Program is to provide a basis
Curtis (Radiation Epidemiology Branch, Epidemiology and Biostatistics Pro-
for assessing progress in reducing the burden of cancer in the
gram. Division of Cancer Etiology), National Cancer Institute, Bethesda, Md.
Correspondence to: Benjamin F. Hankey, ScD., Executive Plaza North, Rm
general population (1).
343J, Rockville, MD 20852.
Journal of the National Cancer Institute Monographs No. 16, 1994
7

---

Fig. 1. Cumulative distribution of breast can-
cer diagnoses by age for the period 1987-1989.
the diagnosis of breast cancer in young women, because it puts
abilities do not differ appreciably by race. The probability of
the occurrence of such an event in a proper perspective. The
developing breast cancer prior to age 50 is .02 and prior to age
method used here to calculate the probability of developing
40 is less than .01.
breast cancer was developed by Feuer et al. (3).
The lifetime probability of developing breast cancer varies Incidence/Mortality Trends
somewhat by race, being .13 in white women, i.e., 13 of 100
Breast cancer incidence trends by age are presented in Fig. 2.
women followed from birth develop breast cancer over their
The log scale is used here and in some subsequent graphs,
lifetime versus .09 in black women. Prior to age 50, these prob- meaning that lines which are Parallel have the same proportional
Fig. 2. Trends in invasive breast cancer in-
cidence rates for following age groups: 20-39,
40-49.50-64, 65+.
8
Journal of the National Cancer Institute Monographs No. 16, 1994

---

rate of change over the X axis, which is calendar year in this
1990. For 1970, 1973 SEER rates were used, since the SEER
case. It is of interest to note that the effect of increased public
Program was not established until 1973. Clearly, there has been
awareness associated with the diagnosis of breast cancer in 1974
a large increase in the number of women diagnosed with breast
in two prominent public figures, Happy Rockefeller and Betty
cancer at ages 20-39. However, the increase in numbers is
Ford, had roughly the same proportional impact in all of the age
primarily due to the increase in the size of the young female
groups, i.e., the lines connecting the 1973 and 1974 rates appear
population, since the incidence rates have remained relatively
to be parallel across all of the age groups, except for the young-
constant over the period (Fig. 2).
est age group where the line appears to be steeper. The trends of
the rates in the three age groups 40 and older have an increasing
Reasons for the Increase in Incidence at Older Ages
slope for the period 1982-1987, with the slope for the oldest age
While the focus here is on breast cancer in younger women, it
group appearing to be the steepest, indicating a faster annual rate of
is of interest to consider possible reasons for the increase in in-
increase as compared to the other two groups. Beginning in
cidence in the older age groups that occurred during the 1980s.
1987, there appears to be a leveling off, if not a decrease, for the
Based on survey data, the rate for the use of mammography for
rates in these age groups. This recent change in the incidence
breast cancer screening, i.e., the percent of women without
trend is not likely due to chance variation.
symptoms who had had a mammogram within the last year, has
During the 1980s up to 1987 when the incidence was rising in
increased from 5% to 10% at the beginning of the decade to
the older age groups, there was little change in the age group 20-
more than 20% in 1987 in women aged 40 and above (4).
39 after a small jump at the beginning of the decade (Fig. 2).
If is assumed that in situ breast cancers, as well as localized
Since 1987 the trend has either not changed or is decreasing.
invasive cancers <2 cm in diameter, are on the average slower
During the conference, a number of individuals indicated that
growing than other breast cancers, then, due to a phenomenon
it was their feeling that there has been an increase in the number
known as length-biased sampling (5), these early cancers should
of young women getting breast cancer because doctors were
be preferentially detected at screening and diagnosed at an ear-
reporting more diagnoses in younger women. Since any increase
lier point in their natural history. The interval between the time
in reported breast cancer diagnoses in younger women was not
at which a cancer is detected at screening and the time at which
likely due to an increase in the incidence rates, it was of interest
it would have been diagnosed clinically is referred to as lead
to assess the change in the size of the young female population
time (5). Over the period that the use of mammography was in-
over time. Fig. 3 presents estimates of the total female popula-
creasing, the incidence of breast cancer in a given calendar year
tion by age for calendar years 1970, 1980, and 1990. It can be
would include all those cancers expected to be diagnosed in the
seen that, over this 20-year period, there has been a dramatic in-
absence of early detection plus those cancers diagnosed early
crease in the number of women in the 20-39-year-old age range.
that would have otherwise been diagnosed at a later time. Even-
Applying the age-specific female breast cancer incidence rates
tually, the increase in incidence should begin to wane because
from the SEER Program to the 5-year age groups in the age
of the number of cancers removed from the total pool of breast
range 20-39 for the three calendar years yields the following
cancers in a given calendar year due to an earlier diagnosis in a
numbers of cancers diagnosed in this age range: 5120 diagnosed
previous calendar year. The dynamics of such a change in in-
in 1970; 7800 diagnosed in 1980: and 10 050 diagnosed in
cidence would be governed by such things as the rate of increase
Fig. 3. Total U.S. female population by age
and calendar year.
Journal of the National Cancer Institute Monographs No. 16, 1994
9

---

in early detection and the amount of lead time associated with
Incidence/Mortality Patterns by Race
the cancers diagnosed early.
There are some interesting contrasts between breast cancer in-
Fig. 4 presents trends in the incidence rates by age for in situ
cidence and mortality patterns for blacks and whites. Mortality
and localized invasive tumors <2 cm in diameter for the period
rates given here are for the total United States. Fig. 5 presents
1983-1989. As expected, rates increase for these two groups in
black/white differences in incidence and mortality by age for the
patients 40 years and older at diagnosis, although only through
period 1987-1989. The incidence of breast cancer is higher in
1987. After 1987, the rates level off. For women 20-39 years of
blacks up to age 40, and breast cancer mortality is higher in
age, there is little change in the rates for localized invasive can-
blacks up to age 65. Differences in incidence and mortality by
cers <2 cm in diameter and a smaller increase in the rate for in
race have existed for several years among younger women (Fig. 6).
situs. Rates for all other invasive cancers (not shown) in all age
groups changed very little during this period. It is of interest to note
Stage at Diagnosis and Survival Patterns by Race
that the incidence of in situ cancers does not increase with age
the way the rate for small invasive cancers does; in fact, there is
The staging system used by the SEER Program to classify
no increase by age in the incidence of in situ cancers above age 50.
cancers consists of the following categories along with their
These observations provide indirect evidence that early detec-
definitions: localized—invasive malignant neoplasm confined
tion played a major role in the increase of breast cancer in-
entirely to the organ of origin: regional—malignant neoplasm
cidence that occurred during the early 1980s. An alternative
that has extended beyond the limits of the organ of origin direct-
possibility is that temporal changes in risk factors selectively af-
ly into surrounding organs or tissues, involves lymph nodes by
fected the incidence of early-stage breast cancers, We are not
way of the lymphatic system, or has both regional involvement
aware of evidence supporting this possibility.
and involvement of regional lymph nodes; distant—malignant
Subsequent to 1987, it is of interest to note that in conjunction
neoplasm that has spread to parts of the body remote from the
with the leveling off of the incidence rates in the older age
primary tumor either by direct extension or by discontinuous
groups, the reported proportion of women who had had a mam-
metastasis (e.g.. implantation or seeding) to distant organs, tis-
mogram within the last 12 months roughly doubled between
sues. or via the lymphatic system to distant lymph nodes: and
1987 and 1990. These data were collected as part of the Nation-
unstaged—insufficient information for staging purposes. This
al Health Interview Survey that is conducted annually by the
system can be applied to patients diagnosed in all years since the
National Center for Health Statistics. Thus, trends in mammog-
beginning of the SEER Program.
raphy utilization and breast cancer incidence during the period
Fig. 7-presents the percent of tumors that were localized to
1987-1990 present something of a conundrum. Efforts to under-
the site of origin at diagnosis by age and race. For breast cancers
stand why breast cancer incidence did not continue to increase
diagnosed in women under the age of 40, there is very little dif-
are currently underway (6). Part of the explanation rests with the
ference in this percent by race or age group. The percent local-
dynamics of the interrelationships between lead time, screening
ized decreases for blacks in the older age groups, which is
rates. and baseline incidence trends as described by Feuer and
somewhat surprising since screening recommendations apply
Wun (7).
primarily to women 50 and older. However, other factors could
Fig. 4. Trends in breast cancer incidence rates
for in situ and localized invasive tumors <=2
cm in diameter for the following age groups:
20-39.40-49.50-64, 65+.
Journal of the National Cancer Institute Monographs No. 16.1994

---

Fig. 5. Age-specific female invasive breast
cancer incidence rates and age-specific female
breast cancer mortality rates (total, United
States) by race.
Fig. 6. Female invasive breast cancer in-
cidence rates and female breast cancer mor-
tality rates (total, United States) for the age
group 20-39 by race and calendar year.
be operating. e. g., at older ages, breast cancers in black women was only little better than that for whites; hoever, the dif-
may be more aggressive than those in whites.
ference was not statistically significant (not shown).
Fig. 8 presents 5-year relative survival rates by race and age
for all stages combined. The rates improve somewhat among
Second Cancer Risk (Incidence)
whites Up to age 50, whereas there is no evidence of an effect of
The risk of subsequent cancers in women with a first invasive
age on survival among blacks. For the age group 20-29, the dif-
breast cancer also varies by age at diagnosis of the first breast
ference between the rates is not statistically significant and there
cancer and, therefore, is considered here. Other reasons for
are no differences in survival by stage (not shown). For women
analyzing second cancer risk associated with female breast can-
30 and older, survival for whites by stage was better than that
cer include 1) obtaining clues about the etiology of selected
for blacks in all cases. except for patients 40-49 years old at
second cancers since, if they occur in excess, this may indicate
diagnosis with distant disease. in which case survival for blacks
the presence of risk factors in common with breast cancer, and
Journal of the National Cancer Institute Monographs No. 16, 1994
11

---

Fig. 7. Percent of invasive breast cancers diag-
nosed as localized by race and age for the
period 1975-1984.
Fig. 8. 5-year relative rates by race and age for
the time period 1975-1984. Relative survival
rates not significantly different in 20-29-year
age group.
2) identifying second cancers that may have been caused by
the analysis. The expected numbers of second cancers were cal-
treatment of the first breast cancer. Clinically, it is of interest to
culated by multiplying cancer site–age–calendar year specific
know which cancers occur in excess in order that patients be
incidence rates, as routinely reported by SEER. by the person-
monitored over time for the occurrence of such malignancies.
years at risk in each cancer site-age-calendar year category. The
Excess risk and relative risk of various second cancers per
numbers were then summed to arrive at the total numbers of ex-
100 000 person-years were calculated. The person-years at risk
pected second cancers.
for a second cancer is defined to be that time starting 2 months after
The size of the first breast cancer cohorts by age were as fol-
diagnosis of the breast cancer and extending to the first of the
lows; <20 (19 women), 20-29 (1431 women), 30-39 (11 002
following events: date of diagnosis of any second cancer, date of
women), 40-49 (26 590 women), and 50+ (127 501 women). A
last follow-up, date of death, or December 31, 1989. Women who
factor limiting this analysis is the fact that the more recently
survived less than 2 months after diagnosis were excluded from
diagnosed cases made a lesser contribution to person-years at
12
Journal of the National Cancer Institute Monographs No. 16, 1994

---

risk of second cancers. since they could only be followed
An excess of second invasive breast cancers is to be expected,
through the end of December 1989.
but the pattern of increased risk in younger women is of interest.
Findings were as follows. Breast cancer patients aged 20-39
In the age group 20-29, the risk of a second invasive breast can-
at diagnosis were at a significantly increased risk of developing a
cer was 29.8 and was related to time from diagnosis of the first
second invasive cancer of the breast, ovary, and lung. Each of
breast cancer (Table 1). Risk decreased from 79.6 during the
these second cancers showed a consistent pattern of being elevated
first year after diagnosis to 15.4 10+ years after diagnosis. In
when the data were analyzed as two cohorts: those patients with
general, the second breast cancer excess decreased with age and
a first invasive breast cancer diagnosed during the period 1973-
time from diagnosis for women with a first breast cancer diag-
1979. and those patients with a first invasive breast cancer diag-
nosed under the age of 50. For women with a first invasive
nosed during the period 1980-1989. Second leukemias were also
breast cancer diagnosed after the age of 50, the relative risk was
found in excess. but have been shown in previous reports (8,9) to be
1.8 and was not related to time from diagnosis.
related to the radiotherapy and chemotherapy given to treat the
Second cancers of the ovary and lung were found to be
initial breast cancer and, therefore, will not be reviewed here.
elevated in women with a first invasive breast cancer diagnosed
Table 1. Risk of second invasive cancer following a first invasive breast cancer diagnosed during the period 1973-1989 by age at diagnosis of the first breast cancer *
Time from diagnosis, y
F i g . 9 .



r i s k

invasive

following


diagnosed during the
period 1973-1989.
Journal of the National Cancer Institute Monographs
16.1994
13

---

at ages 30-49 (Table 1). The risk of a second ovarian cancer was
(2) American Cancer Society, Inc.: Cancer Facts & Figures-1993, 1993
(3) Feuer EJ, Wun LM, Boring CC, et al: The lifetime risk of developing
highest in the age group 30-39. The relative risks for both can-
breast cancer. J Natl Cancer Inst 85:848-849, 1993
cers were not related to time from diagnosis. The finding of
(4) Sondik EJ.,Kessler LG, Ries LAG, eds: Cancer Statistics Review: 1973-
elevated risks for second breast and ovarian cancers in young
1986, including a report on the status of cancer control. National Cancer
Institute. NIH Publ. No. 89-2789, 1989, pp II.20-II.22
women with a first breast cancer is consistent with the iden-
(5) Zelen M: Theory of early detection of breast cancer in the general popula-
tification of a genetic marker, which has been reported to be as-
tion. In Breast Cancer: Trends in Research and Treatment (Heusen JC,
sociated with the occurrence of breast and ovarian cancers at a
Mattheiem WH, Rozencweig M, eds). New York: Raven Press, 1976, pp
287-300
very young age (10).
(6) Kessler LG. Breen N: Annual use of screening mammography doubles for
Excess risk (absolute risk) of developing a second invasive
women 40 and over in the U. S. between 1987 and 1990: Evidence from
breast cancer was calculated by age at diagnosis of the first in-
the National Health Interview Surveys. Submitted for publication
Feuer EJ, Wun LM: How much of the recent rise in breast cancer in-
vasive breast cancer (Fig. 9). This risk was quite high in the 20-
(7)
cidence can be explained by increases in mammography utilization? Am J
29 age group (893.8 per 100 000 person years at risk). The rate
Epidemiol 136: 1423-1436, 1992
decreased with increasing age at diagnosis of the first breast
(8) Curtis RE, Boice JD Jr, Moloney MC, et al: Leukemia after chemotherapy
for breast cancer. Cancer Res 50:2741-2746, 1990
cancer, but remained high relative to the expected rate.
(9) Curtis RE, Boice JD Jr, Stovall LM, et al: Risk of leukemia after chemo-
therapy and radiation treatment for breast cancer. N Engl J Med 326: 1745-
1751, 1992
References
(10) Hall JM. Friedman L, Guenther C. et al: Closing in on a breast cancer gene
on chromosome 17q. Am J Hum Genet 50: 1235-1242, 1992
(1) Miller BA, Ries LAG, Hankey BF, et al, eds: Cancer Statistics Review:
1973-1989. National Cancer Institute. NIH Publ. No. 92-2789.1992
14
Journal of the National Cancer Institute Monographs No. 16.1994

---

Document Outline

• Data Sources
• Results
  • Proportion of Breast Cancers Diagnosed at Younger Ages
  • Probability of Developing Breast Cancer
  • Affiliations of authors:
  • Correspondence to:
  • Incidence/Mortality Trends
  • Reasons for the Increase in Incidence at Older Ages
  • Incidence/Mortality Patterns by Race
  • Stage at Diagnosis and Survival Patterns by Race
  • Second Cancer Risk (Incidence)
  • References

---