Growth of uterine leiomyomata among premenopausal black and white women
Uterine leiomyomata (fibroids) are the leading cause of hysterectomy in the United States. Black women have a greater fibroid burden than whites, yet no study has systematically evaluated the growth of fibroids in blacks and whites. We prospectively tracked growth for 262 fibroids (size range: 1–13 cm in diameter) from 72 premenopausal participants (38 blacks and 34 whites). Fibroid volume was measured by computerized analysis of up to four MRI scans over 12 months. We used mixed effects models to identify factors that are associated with growth, and results were converted to percent change per 6 months for clinical relevance. The median growth rate was 9% (range: −89% to +138%). Seven percent of fibroids regressed (>20% shrinkage). Tumors from the same woman grew at different rates (within-woman component of variation was twice the component among women; both were significant, P < 0.001). Black and white women less than 35 years of age had similar fibroid growth rates. However, growth rates declined with age for whites but not for blacks (P = 0.05). The odds of a tumor growing more than 20% in 6 months also decreased with age for whites but not for blacks (P < 0.01). Growth rates were not influenced by tumor size, location, body mass index, or parity. We conclude that (i) spontaneous regression of fibroids occurs; (ii) fibroids from the same woman grow at different rates, despite a uniform hormonal milieu; (iii) fibroid size does not predict growth rate; and (iv) age-related differences in fibroid growth between blacks and whites may contribute to the higher symptom burden for black women. • ethnic • fibroid • MRI • tumor growth • longitudinal data
Growth of uterine leiomyomata among
premenopausal black and white women
Shyamal D. Peddada
, Shannon K. Laughlin
, Kelly Miner
, Jean-Philippe Guyon
, Karen Haneke
Heather L. Vahdat
, Richard C. Semelka
, Ania Kowalik
, Diane Armao
, Barbara Davis
, and Donna Day Baird
aBiostatistics Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC 27709; bEpidemiology Branch, National Institute of
Environmental Health Sciences, Research Triangle Park, NC 27709; cIntegrated Laboratory Systems, Durham, NC 27713; dDepartment of Radiology, University
of North Carolina, Chapel Hill, NC 27599; eDepartment of Obstetrics/Gynecology, University of North Carolina, Chapel Hill, NC 27599; and fLaboratory of
Women’s Health, National Institute of Environmental Health Sciences, Research Triangle Park, NC 27709
Edited by R. Michael Roberts, University of Missouri, Columbia, MO, and approved October 16, 2008 (received for review August 18, 2008)
Uterine leiomyomata (ﬁbroids) are the leading cause of hysterec-
tomy in the United States. Black women have a greater ﬁbroid
burden than whites, yet no study has systematically evaluated the
growth of ﬁbroids in blacks and whites. We prospectively tracked
growth for 262 ﬁbroids (size range: 1–13 cm in diameter) from 72
premenopausal participants (38 blacks and 34 whites). Fibroid
volume was measured by computerized analysis of up to four MRI
scans over 12 months. We used mixed effects models to identify
factors that are associated with growth, and results were con-
verted to percent change per 6 months for clinical relevance. The
median growth rate was 9% (range: ⴚ89% to ⴙ138%). Seven
percent of ﬁbroids regressed (>20% shrinkage). Tumors from the
same woman grew at different rates (within-woman component of
variation was twice the component among women; both were
signiﬁcant, P<0.001). Black and white women less than 35 years
of age had similar ﬁbroid growth rates. However, growth rates
declined with age for whites but not for blacks (Pⴝ0.05). The odds
of a tumor growing more than 20% in 6 months also decreased
with age for whites but not for blacks (P<0.01). Growth rates were
not inﬂuenced by tumor size, location, body mass index, or parity.
We conclude that (i) spontaneous regression of ﬁbroids occurs; (ii)
ﬁbroids from the same woman grow at different rates, despite a
uniform hormonal milieu; (iii) ﬁbroid size does not predict growth
rate; and (iv) age-related differences in ﬁbroid growth between
blacks and whites may contribute to the higher symptom burden
for black women.
ethnic 兩ﬁbroid 兩MRI 兩tumor growth 兩longitudinal data
Uterine leiomyomata (fibroids) are the leading indication for
hysterectomy in the United States (1). Myomectomy and
uterine artery embolization are also common treatments, but
hysterectomy may be required subsequently (2). Hartmann et al.
(3) estimate a $4,600 excess health care cost during the year
following each US woman’s diagnosis of fibroids. National
medical costs associated with fibroids exceed 2 billion dollars
annually (4). African Americans have a higher fibroid incidence
(5, 6), experience more severe symptoms (7), present with larger
tumors (7), and have a threefold higher risk of hysterectomy (8)
compared with whites. Symptoms increase with the size of
fibroids (7, 9, 10). However, few studies have examined the
growth of fibroids over time (11–13), and no study has system-
atically followed the growth of fibroids in black and white
The Fibroid Growth Study was designed to measure the
growth of fibroids in black and white women with clinically
relevant fibroids using MRI technology. We compare growth
rates of individual tumors from the same woman; contrast fibroid
growth in black and white women; and examine associations with
age, parity, body mass index (BMI), and tumor characteristics.
Study Participants. Characteristics of the 72 participants are
shown in Table 1. Our cohort ranged in age from 24 to 54 years,
and approximately half were African American. Nearly 60%
were overweight or obese. More than half were diagnosed with
fibroids within 2 years of study entry. Most had multiple fibroids,
and approximately a third had more than eight. As expected
from the criteria for entry, participants had enlarged uteri
(range: 110–1,995 cm
, with nearly a fifth greater than 1,000
). Fifteen of the 72 women opted for treatment during the
study year. There were no statistically significant differences
between black and white women with respect to these charac-
teristics, but there was a tendency for blacks to be younger and
to have higher BMIs. Women were recruited irrespective of
symptoms. Although most (88%) reported problems with pelvic
pain or bleeding, only 31% reported that these symptoms limited
Fibroid Characteristics at Enrollment. Fibroid size, type, and loca-
tion are shown in Table 1. The initial volume of the 262 measured
fibroids varied from 1.3 to 1098 cm
with a median volume of 17.3 cm
(3.2-cm diameter). The 6
submucosal fibroids tended to be small, with a median size of 7.7
Fibroid and Uterine Growth Rates. The 262 tumors varied widely in
their growth rates; they ranged from shrinkage of 89% to growth
of 138% per 6 months (Fig. 1). The median fibroid growth rate
for both black and white women was 9% per 6 months. Eighty-
eight tumors (34%) were rapidly growing (⬎20% increase in
volume per 6 months), and 19 (7%) were spontaneously regress-
ing (⬎20% decrease in volume per 6 months).
We next wanted to determine how fibroid growth was related
to overall uterine growth. The median uterine growth rate was
6% per 6 months. Despite our only measuring a subset of fibroids
for most women, those who had at least one measured fibroid
that was rapidly growing had significantly increased uterine
growth compared with those without a rapidly growing tumor
(P⫽0.027). Uterine growth rate averaged 14% higher for
women with a measured rapidly growing tumor compared with
Author contributions: S.D.P., K.H., R.S., A.K., D.A., B.D., and D.B. designed research; S.D.P.,
S.L., K.M., J-P.G., K.H., H.L.V., R.S., D.A., B.D., and D.B. performed research; S.D.P., S.L., K.M.,
J-P.G., H.V., and D.B. analyzed data; and S.D.P., S.L., B.D., and D.B. wrote the paper.
The authors declare no conﬂict of interest.
This article is a PNAS Direct Submission.
1Present address: Novartis, Cambridge, MA 02139.
2Present address: PMB 347, 4093 Diamond Ruby, Suite 7, Christiansted, Virgin Islands 00820.
3Present address: Family Health International, Durham, NC 27713.
4Present address: Reproductive Science Center, Lexington, MA 02421.
5Present address: Millennium: The Takeda Oncology Company, Cambridge, MA 02139.
6To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
December 16, 2008
Spontaneously Regressing Tumors. Nineteen tumors from 14
women showed spontaneous shrinkage. When examined descrip-
tively, they were found to vary in size and location and to come
from both blacks and whites aged 30–49 years [see supporting
information (SI) Table S1]. We looked at the MRI scans showing
these 19 tumors for lack of gadolinium enhancement, which
would be consistent with loss of arterial blood f low and necrosis
(14). By visual estimate, 7 tumors showed necrosis exceeding
50% of tumor volume and 2 others showed 20% to 40% necrosis.
Tumors with more dramatic shrinkage tended to have greater
Table 1. Characteristics of the participants (nⴝ72) and their ﬁbroids (nⴝ262), Fibroid Growth Study, enrollment 2001–2004
⬍35 23 31.9 14 36.8 9 26.5
35–44 28 38.9 17 44.7 11 32.3
ⱖ45 21 29.2 7 18.5 14 41.2
0 43 59.7 21 55.3 22 64.7
ⱖ1 29 40.3 17 44.7 12 35.3
⬍25 30 41.6 12 31.6 18 53.0
25–29.9 21 29.2 13 34.2 8 23.5
ⱖ30 21 29.2 13 34.2 8 23.5
Time since initial diagnosis of ﬁbroids (yr)
⬍1 25 34.7 10 26.3 15 44.1
1–2 17 23.6 10 26.3 7 20.6
3–5 8 11.1 6 15.8 2 5.9
5–9 12 16.7 7 18.4 5 14.7
ⱖ10 8 11.1 5 13.2 3 8.8
missing 2 2.8 0 0.0 2 5.9
Uterine volume, cm
⬍250 14 19.4 7 18.4 7 20.6
250–499 19 26.4 10 26.3 9 26.5
500–999 26 36.1 14 36.8 12 35.3
ⱖ1,000 13 18.1 7 18.4 6 17.6
Number of ﬁbroids
1 5 6.9 1 2.6 4 11.8
2 11 15.3 5 13.2 6 17.6
3–8 27 37.5 15 39.5 12 35.3
⬎8 29 40.3 17 44.7 12 35.3
None 57 79.1 28 73.7 29 85.3
Embolization 2 2.8 1 2.6 1 2.9
Myomectomy 7 9.7 6 15.8 1 2.9
Hysterectomy 6 8.3 3 7.9 3 8.8
None 52 72.2 27 71.1 25 73.5
Oral contraceptives 16 22.2 8 21.0 8 23.5
Other 4 5.6 3 7.9 1 3.0
Initial ﬁbroid volume (diameter
(⬍3.0 cm) 121 46.1 75 48.5 46 43.0
(3.0–4.9 cm) 82 31.3 52 33.6 30 28.0
(ⱖ5.0 cm) 59 22.5 28 18.1 31 29.0
Submucosal 6 2.3 6 3.9 0 0.0
Intramural 166 63.4 99 63.9 67 62.2
Subserosal 90 34.4 50 32.3 40 37.4
Corpus 147 56.1 85 54.8 62 57.9
Fundus 56 21.4 32 20.6 24 22.4
Lower segment 59 22.5 38 24.5 21 19.6
*Treated women had growth data censored at treatment (5 women had 4 MRI scans, but 3 were censored after 3 MRI scans, and 7 were censored after 2 MRI
†Diameter calculated from measured volume based on ellipsoid formula.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808188105 Peddada et al.
necrosis (Spearman correlation ⫽⫺0.56, P⫽0.013; Table S1).
In a comparison sample of 19 tumors randomly selected from the
remaining 243, 1 tumor showed 20% necrosis and none showed
greater than 20% necrosis.
Fibroids from the Same Woman Grow at Different Rates. Growth
rates for each woman’s tumors are shown in Fig. 2, with the 72
women ordered by the median growth rate of their tumors. The
individual tumor growth rates are represented by the hatch
marks on the vertical lines. These demonstrate the wide range of
fibroid growth rates within a given woman. Seven women had
both a rapidly growing and a spontaneously regressing tumor.
Fig. 2 also shows that despite the within-woman variability,
some women tend to have tumors that grow rapidly, whereas
other women tend to have stable or shrinking tumors. When the
total variation in tumor growth rate was partitioned into a
component for within-woman variation and a component for
between-women variation using mixed model regression, both
components were significant (P⬍0.001). The variation of tumor
growth within women was two times the variation between
Factors Related to Rate of Fibroid Growth. Fig. 3 shows the asso-
ciations of fibroid growth with several characteristics, including
ethnicity, age, number of fibroids, and size of the fibroid. The
results are based on analysis of 258 fibroids from 72 women after
excluding statistical outliers (the four most rapidly shrinking
tumors; see SI Text for description of effects of the outliers). The
mean tumor growth rate for blacks in the study was similar to that
for whites (12% vs. 10% increase in volume per 6 months,
respectively). However, when we compared tumors within age
categories by ethnicity, the tumors from older white women (ⱖ45
years) grew much more slowly than those from older black
women (2% vs. 15% growth rate, respectively; Fig. 3). Adjusted
analyses based on linear mixed effects models supported this
association, demonstrating a significant decline in growth rate
with age for tumors from whites but not from blacks (P⫽0.05;
Table 2). The pair-wise analysis shows a significant decrease in
tumor growth rate for older whites when compared with younger
whites (P⫽0.004), and no such difference was found among
blacks (P⫽0.67). Furthermore, the chance of a tumor growing
rapidly (⬎20% increase in volume per 6 months) depended on
age for white women but not for black women (P⫽0.004). The
relative odds of rapid growth for younger whites was 17 times
that of the older whites (P⬍0.001). For blacks there was no
significant difference by age (P⫽0.81). These results are
summarized in Table S2.
The only other factor affecting fibroid growth rate was the
number of fibroids in the uterus. Single tumors grew much faster
than fibroids that shared a uterus (Fig. 3, Table 2). Fibroid
growth rates were not significantly associated with BMI or
parity, or with tumor size, type, or location. The reader is
referred to SI Text for sensitivity analyses (including dropping
the six submucosal fibroids and dropping the five women with a
single fibroid) that demonstrated robustness of results.
With the exception of the factors that we found to be
important (age, ethnicity, and number of tumors), other char-
acteristics were associated with ⬍5% difference in growth rates.
Although such differences would be statistically significant with
extremely large sample sizes, they would likely not be viewed as
Fig. 1. Frequency distribution of growth rates for 262 ﬁbroids from 72
premenopausal women, Fibroid Growth Study, enrollment 2001–2004.
Fig. 2. Median (black circles) and range (vertical bars) of ﬁbroid growth rates
for each of 72 participants, ordered by each participant’s median tumor
growth rate. The horizontal hatch marks show the growth rate for each
individual tumor, Fibroid Growth Study, enrollment 2001–2004.
Fig. 3. Unadjusted mean ﬁbroid growth rates by participant characteristics
and ﬁbroid characteristics, 258 ﬁbroids (4 statistical outliers excluded) from 72
participants, Fibroid Growth Study, enrollment 2001–2004.
Peddada et al. PNAS
December 16, 2008
clinically important. Using our variance estimates, we found that
our study had 80% power to detect differences in growth rates
This longitudinal study of 262 uterine leiomyomata in 72 pre-
menopausal women provides an in-depth analysis of tumor
growth in black and white women. We demonstrated that
fibroids within the same woman often have different growth
rates despite having a similar hormonal milieu. Indeed, each
tumor appeared to have its own intrinsic growth rate, consistent
with studies showing that fibroids are monoclonal in origin (15,
16) with variable molecular characteristics (17–21). The only
other large study of fibroid growth was conducted in Japan, and
most of the 70 participants had only a single tumor (11); thus,
variation in growth of fibroids from the same woman could not
We observed spontaneous regression in a small percentage of
fibroids, surprising in premenopausal women. Tumor shrinkage
after menopause is assumed to occur; there is a dramatic
reduction in clinical diagnoses after menopause (22), and post-
menopausal fibroids are predominantly small lesions (23). De-
waay et al. (12) identified six small spontaneously resolved
tumors in women approaching menopause; otherwise, however,
spontaneously regressing tumors in premenopausal women have
not been well documented in the literature. The women with
regressing tumors in our study were having regular menses, and
half were in their 30s. Many of the shrinking tumors we observed
showed evidence of necrosis, suggesting that vascular events may
A fundamental question we sought to address is whether
fibroid growth differs in black and white women. There is a
general assumption that fibroids grow faster in black women
compared with white women because black women are diag-
nosed at a younger age and have a higher incidence and more
symptoms (5–8). Molecular markers also may differ between
tumors from blacks and whites (18, 21, 24). We found significant
ethnic differences in fibroid growth when age was considered.
Growth rates were similar between blacks and whites in the
youngest age group (⬍35 years) but declined in older white
women so that, on average, tumors grew extremely slowly in
white women in the ⱖ45 age group. In contrast, growth rates
showed little decline with increasing age for blacks. Importantly,
the greater fibroid burden observed in black women may be
explained by our observation that fibroid growth rates show little
decline with increasing age in black women and by the previously
reported finding that black women have an earlier onset com-
pared with white women (5, 6, 25).
The decline in tumor growth rate in older white women was
an unexpected finding and could not be attributed to these
women entering menopause. Participants were having regular
periods, and the decline was not seen exclusively in the oldest age
group. Instead, there was a gradual decline across the three age
groups for whites. Even if some were perimenopausal, their
fibroids would have been expected to continue growing based on
the clinical literature (26, 27), which refers to perimenopausal
instability of ovarian function and more frequent periods of
unopposed estrogen as a possible mechanism for rapid growth (27).
Dysregulation of the extracellular matrix has been suggested as
an important etiologic factor in fibroid growth (28). There may
be age-related changes in angiogenesis or extracellular matrix
production that differ between blacks and whites. Future studies
should consider the age of the woman when looking at fibroid
The finding that fibroid growth was not inf luenced by tumor
characteristics such as size and location was surprising and is
important for research that characterizes molecular character-
istics of fibroid tissue. Tumor size has been related to variation
in molecular markers (17, 19–21), and it has been assumed that
the molecular differences reflect differences in tumor growth
rates. Our data show that large tumor size cannot be used as an
indicator of a growing tumor.
Our data showed more rapid growth for solitary tumors than
for multiple tumors that share a uterus. However, our sample
size of women with solitar y tumors was quite small, and we
required participants to have either at least one large fibroid or
an enlarged uterus for enrollment. It is possible that only rapidly
growing tumors can attain a large size while remaining solitary,
so the finding could be attributable to our sample selection.
Alternatively, solitary tumors may grow faster because of less
competition for uterine blood supply. To evaluate these alter-
natives, small solitary tumors need to be studied.
The study has other limitations. Extremely small tumors
(⬍1.5-cm diameter) could not be measured accurately, and our
focus on women with at least one already well-developed fibroid
Table 2. Adjusted growth rate differences* associated with
participant and ﬁbroid characteristics, Fibroid Growth Study,
Adjusted differences in
growth rate estimate
(95% conﬁdence interval)
Age by ethnicity, yr 0.05
Blacks ⬍35 Reference
Blacks 35–44 ⫺7.05 (⫺17.62, 4.87)
Blacks ⱖ45 ⫺3.38 (⫺17.31, 12.90)
Whites ⬍35 Reference
Whites 35–44 ⫺7.76 (⫺21.04, 7.74)
Whites ⱖ45 ⫺19.57
Number of ﬁbroids 0.06
3–8 5.69 (⫺2.89, 42.32)
2 0.72 (⫺13.73, 43.26)
1 33.75 (7.58, 61.17)
25–29.9 ⫺4.81 (⫺13.79, 38.67)
ⱖ30 2.40 (⫺7.30, 41.61)
ⱖ1 4.17 (⫺5.86, 42.41)
Initial ﬁbroid volume
(⬍3 cm) Reference
(3.0–4.9 cm) ⫺3.53 (⫺9.43. 37.80)
(ⱖ5.0 cm) ⫺3.10 (⫺9.96, 38.36)
Subserosal 0.55 (⫺5.12, 39.20)
Fibroid location 0.64
Fundus ⫺3.25 (⫺9.67, 38.12)
Lower segment ⫺0.75 (⫺7.49, 39.17)
n⫽258 ﬁbroids from 72 women. Four statistical outliers (all shrinking
⬎50% in volume per 6 months) were excluded, leaving 258 ﬁbroids.
*Age by ethnicity differences are adjusted for number of ﬁbroids; number of
ﬁbroid differences are adjusted for age by ethnicity; all other variables are
adjusted for age by ethnicity and number of ﬁbroids.
†Pvalue for the overall importance of each factor for growth.
‡Pair-wise difference between ⱖ45 and ⬍35 group is statistically signiﬁcant at
§Diameter calculated from measured volume based on ellipsoid formula.
¶Intramural group includes six submucosal ﬁbroids.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808188105 Peddada et al.
did not allow us to examine initial tumor development. We also
were unable to examine submucosal fibroids statistically because
there were so few of them. However, we investigated potential
biases, including biased sample selection, variation in menstrual
phase at time of MRI, and other possible confounders and found
little effect on our findings (SI Materials).
Our findings have implications for patient care and for further
research directions. Current clinical practice encourages an
ultrasound or pelvic examination at 6 months to evaluate growth
(29). Our analysis shows that the majority of tumors grow less
than 20% in 6 months, with a median growth rate of 9%. Thus,
it may be possible to extend the follow-up time for clinical
assessment of fibroid growth. In addition, if further research
supports our findings that tumor growth rates decline in white
women as they age, those approaching perimenopause might
choose to delay treatment and wait for menopause when tumors
are likely to shrink. Current medical therapies have focused on
hormonal manipulation of well-developed tumors (30). The
rapid growth of tumors in young women in both ethnic groups
suggests that research is urgently needed to study tumor onset
and identify preventive factors. Treatments that inhibit early
tumor growth could stop development of debilitating symptoms.
Study Design. The National Institute of Environmental Health Sciences Fibroid
Growth Study was a collaborative study with the University of North Carolina
Medical Center that enrolled participants from 2001 to 2004 with approval
from both institutional review boards. Premenopausal women with a known
diagnosis of ﬁbroids, conﬁrmed by ultrasound, were recruited from gynecol-
ogy clinics and announcements in the community (Davis et al., in review). To
ensure clinical relevance, enrollment was limited to women with at least one
ﬁbroid greater than 5 cm in diameter or a uterus enlarged to at least a 12-week
pregnancy size (200–250 cm3) (31). Fibroids were measured up to four times
(MRI scans taken at enrollment and then at 3, 6, and 12 months). Of the 116
participants, 35 completed only one MRI scan (30 women opted for treatment,
4 women dropped out, and 1 woman had completed only one MRI scan when
the ﬁeld study ended). Of the 81 women with two or more MRI scans, 3 were
excluded because their tumors were smaller than the size criteria for this
analysis. We further limited analysis to black and white women. This left 72
women in our analysis sample (38 blacks and 34 whites). Prospective time in
the study averaged approximately 9 months, primarily because of early ter-
mination when the ﬁeld study ended (see details in SI Text).
Measurement of Tumor and Uterine Volume. Sagittal T2-weighted MRI scans
without contrast were evaluated for type (submucosal, intramural, or subse-
rosal), location, and size of ﬁbroids (see detailed description of MRI protocol
in SI Text). Fibroids were selected for volumetric measurement if they were
seen in at least three consecutive slices and had traceable borders. When a
woman had many ﬁbroids, the technician selected tumors representing dif-
ferent sizes and positions in the uterus. All submucosal ﬁbroids were mea-
sured. The ﬁnal sample was limited to tumors that had volumes ⬎5.0 cm3or
were seen on at least ﬁve consecutive slice images (n⫽262 ﬁbroids, with each
woman contributing 1–11 individual tumors).
Fibroid volume was determined using the volume estimation and tracking
over time method developed for this study (32). All analyses used volumetric
measures, but a diameter size was calculated from measured volume for
descriptive purposes. Uterine volume was estimated from each participant’s
ﬁrst and last MRI scan based on measurement of craniocaudal length (L),
transverse width (W), and anterior/posterior (AP) diameter, and application of
the ellipsoid formula (L ⫻W⫻AP ⫻0.52). Details of measurement and quality
assurance are in SI Text.
Tumor Type and Location. The type of ﬁbroid was deﬁned by the position of its
center in relation to the inner and outer boundaries of the uterus. Submucosal
tumors were centered in the endometrial lining, intramural in the myome-
trium, and subserosal along the external lining or outside of the uterus. The
location of a ﬁbroid within the uterus was deﬁned by the position of its center
in relation to the fundus, corpus, or lower uterine segment (see reference
diagram in SI Text).
Determination of Fibroid and Uterine Growth Rates. The natural logarithm of
ﬁbroid volume was used to make the distribution approximately normal. The
growth rate of each tumor was based on the change in log volume between
each MRI scan divided by the number of days in the interval. For each tumor,
rates across intervals were averaged. For clearer clinical application, the
average growth rates were converted to a 6-month percent change in volume.
A 6-month interval was chosen because that is a clinically recommended
follow-up period (29) and it falls within the observation period of this study.
The reader is referred to SI Text for further details of growth rate determi-
nations. Uterine growth rate was deﬁned as the change in log volume divided
by the number of days between the ﬁrst and last MRI scans, summarized as
percent change in volume per 6 months.
Statistical Analyses. We used mixed effects linear regression models to eval-
uate factors that may inﬂuence ﬁbroid growth rates (PROC MIXED, version
9.1; SAS Institute, Cary, NC). This method accounted for any correlation in
growth rates among tumors from the same woman. Details of the analysis and
model selection are in SI Text. We investigated ethnicity (blacks vs. whites,
based on self-reported ethnicity), age, number of ﬁbroids in the uterus at
study enrollment (1, 2, 3– 8, or ⬎8), participant BMI (kg/m3based on measure-
ments taken at the ﬁrst MRI scan), participant parity (nulliparous vs. parous
based on self-report), initial ﬁbroid volume, ﬁbroid type (subserosal vs. the
combined group of intramural and submucosal, combined because there were
only 6 submucosal ﬁbroids), and location of ﬁbroid (fundus, corpus, or lower
segment) as potential factors associated with tumor growth. Statistical sig-
niﬁcance of pair-wise differences was evaluated only when a factor was
signiﬁcant at P⫽0.05. We also investigated the factors associated with the
odds of a tumor growing rapidly (20% or more growth in a 6-month period)
using random effects logistic regression analysis (SAS macro GLIMMIX; SAS
Institute, Cary, NC) (see SI Text for details). We conducted analyses with the
same primary and secondary variables as in the growth rate model and used
the same 258 tumors in analysis.
ACKNOWLEDGMENTS. The authors thank study participants; Ms. Edelstein
(National Institute on Environmental and Health Sciences [NIEHS]) for graph-
ics; Ms. Greasby (Graduate Student, Department of Biostatistics, University of
California, Davis) for initial data quality control checks; Ms. Firat (Radiology
Technician, University of North Carolina, Chapel Hill); Mss. Daniel, Davis, and
Pope for managing radiology ﬁlms and digital images (all at Integrated
Laboratory Systems); and Ms. Ragan and Drs. Kissling, Korach, Weinberg, and
Wilcox (NIEHS) and Dr. Leppert (Duke University) for reviewing an earlier
version of the manuscript. Research was supported by the Intramural Research
Program of the NIEHS, National Institutes of Health (Grant Z01ES 101663-05),
with partial funding from the National Center on Minority Health and Health
Disparities Grant MO1RR00046 and NIEHS Contracts N01-ES-95446 and 273-
1. Farquhar CM, Steiner CA (2002) Hysterectomy rates in the United States 1990–1997.
Obstet Gynecol 99:229–234.
2. Edwards RD et al. (2007) Committee of the Randomized Trial of Embolization versus
Surgical Treatment for Fibroids. Uterine-artery embolization versus surgery for symp-
tomatic uterine ﬁbroids. N Engl J Med 356:360–370.
3. Hartmann KE, et al. (2006) Annual costs associated with diagnosis of uterine leiomy-
omata. Obstet Gynecol 108:930–937.
4. Flynn M, Jamison M, Datta S, Myers E (2006) Health care resource use for uterine ﬁbroid
tumors in the United States. Am J Obstet Gynecol 195:955–964.
5. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM (2003) High incidence of
uterine leiomyoma in black and white women: Ultrasound evidence. Am J Obstet
6. Marshall LM, et al. (1997) Variation in the incidence of uterine leiomyoma among
premenopausal women by age and race. Obstet Gynecol 90:967–973.
7. Kjerulff KH, Langenberg P, Seidman JD, Stolley PD, Guzinski GM (1996) Uterine leiomyomas.
Racial differences in severity, symptoms and age at diagnosis. J Reprod Med 41:483–490.
8. Myers ER, et al. (2001) Management of uterine ﬁbroids (Evidence Report/Technology
Assessment No. 34, Contract 290–97-0014 to the Duke Evidence-Based Practice Center).
AHRQ Publication No. 01-E052 (Agency for Healthcare Research and Quality, Rockville,
9. Wegienka G, et al. (2003) Self-reported heavy bleeding associated with uterine leiomy-
omata. Obstet Gynecol 101:431–437.
10. Vergani P, et al. (2007) Large uterine leiomyomata and risk of cesarean delivery. Obstet
Gynecol 109:410– 414.
11. Tsuda H, Kawabata M, Nakamoto O, Yamamoto K (1998) Clinical predictors in the
natural history of uterine leiomyoma: preliminary study. J Ultrasound Med 17:17–
12. DeWaay DJ, Syrop CH, Nygaard IE, Davis WA, Van Voorhis BJ (2002) Natural history of
uterine polyps and leiomyamata. Obstet Gynecol 100:3–7.
13. Ichimura T, et al. (1998) Correlation between the growth of uterine leiomyomata and
estrogen and progesterone receptor content in needle biopsy specimens. Fertil Steril
Peddada et al. PNAS
December 16, 2008
14. Shimada K, Ohashi I, Kasahara I, Miyasaka N, Shibuya H (2004) Triple-phase dynamic
MRI of intratumoral vessel density and hyalinization grade in uterine leiomyomas. AJR
15. Hashimoto K et al. (1995) Clonal determination of uterine leiomyomas by analyzing
differential inactivation of the X-chromosome-linked phosphoglycerokinasegene. Gy-
necol Obstet Invest 40:204–208.
16. Cai YR, et al. (2007) X-chromosomal inactivation analysis of uterine leiomyomas reveals
a common clonal origin of different tumor nodules in some multiple leiomyomas. Int
J Oncol 31:1379–1389.
17. Lobel MK, Somasundaram P, Morton CC (2006) The genetic heterogeneity of uterine
leiomyomata. Obstet Gynecol Clin North Am 33:13–39.
18. Wei J-J, et al. (2006) Ethnic differences in expression of the dysregulated proteins in
uterine leiomyomata. Hum Reprod 21:57–67.
19. Wei J-J, Chiriboga L, Mittal K (2005) Expression proﬁle of the tumorigenic factors
associated with tumor size and sex steroid hormone status in uterine leiomyomata.
Fertil Steril 84:474– 484.
20. Wolanska M, Bankowski E (2007) Transforming growth factor beta and platelet-
derived growth factor in human myometrium and in uterine leiomyomas at various
stages of tumour growth. Eur J Obstet Gynecol Reprod Biol 130:238–244.
21. Wang T, et al. (2007) A micro-RNA signature associated with race, tumor size, and target
gene activity in human uterine leiomyomas. Genes Chromosomes Cancer 46:336–347.
22. Ross RK, et al. (1986) Risk factors for uterine ﬁbroids: reduced risk associated with oral
contraceptives. Br Med J 293:359–362.
23. Cramer SF, Patel A (1990) The frequency of uterine leiomyomas. Am J Clin Pathol
24. Pan Q, Luo X, Chegini N (2007) Genomic and proteomic proﬁling I: Leiomyomas in
African Americans and Caucasians. Reprod Biol Endocrinol 5:34– 48.
25. Wise LA. Palmer JR, Stewart EA, Rosenberg L (2005) Age-speciﬁc incidence rates for
self-reported uterine leiomyomata in the Black Women’s Health Study. Obstet Gynecol
26. Ryan GL, Syrop CH, VanVoorhis BJ (2005) Role, epidemiology, and natural history of
benign uterine mass lesions. Clin Obstet Gynecol 48:312–324.
27. Wallach EE, Vlahos NF (2004) Uterine myomas: An overview of development, clinical
features, and management. Obstet Gynecol 104:393–406.
28. Leppert PC, Catherino WH, Segars JH (2006) A new hypothesis about the origin of
uterine ﬁbroids based on gene expression proﬁling with microarrays. Am J Obstet
29. Katz V, Lentz G (2007) Congenital abnormalities of the female reproductive tract:
Anomalies of the vagina, cervix, uterus, and adnexa. Comprehensive Gynecology, eds
Katz V, Lentz G, Lobo R, Gershenson D (Mosby, St. Louis), 5th Ed.
30. Stewart EA (2001) Uterine ﬁbroids. Lancet 357:293–298.
31. Harb TS, Adam RA (2005) Predicting uterine weight before hysterectomy: Ultrasound
measurements versus clinical assessment. Am J Obstet Gynecol 193:2122–2125.
32. Guyon J-P, et al. (2003) VETOT, volume estimation and tracking over time: Framework
and validation. Lecture Notes Comput Sci 2879(Pt 2):142–149.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0808188105 Peddada et al.