Self-reported family history of leiomyoma: not a reliable marker of high risk

Abstract

Purpose

To examine the importance of self-reported family history of uterine leiomyoma (fibroids) as a marker of risk.

Methods

Women, aged 35-49, were randomly selected from the membership of a large urban health plan. Participants completed a self-administered questionnaire about family history of fibroids. Ultrasound screening for fibroids followed, regardless of whether participants had been previously diagnosed (660 black, 412 white). Data for each ethnic group were analyzed separately using Poisson regression.

Results

In both ethnic groups women who reported a family history of fibroids had an elevated risk of fibroids compared to those without family history. However, no elevated risk was apparent for cases who did not know they had fibroids when they reported the family history information.

Conclusions

Many women may first learn about their family history of fibroids when discussing their own clinical diagnosis with family members. Such bias would invalidate self-reported family history as a predictor of fibroid risk. As new pharmacological treatments for fibroids are developed, women at high-risk of fibroids would benefit from early screening and pharmacologic-treatment to delay development of large fibroids and reduce the need for invasive treatments. Self-reported family history is not useful for identifying high risk women.

Introduction

Fibroids are the most common reproductive tract tumor in women, and the leading indication for hysterectomy in the United States [1, 2]. Fibroids contribute to many common gynecologic problems including pelvic pain and menorrhagia, the latter of which can lead to life threatening anemia [3]. U.S. medical costs associated with this condition are estimated at $4 to $10 billion annually [4]. The tumors are clonal in growth [5], and a minority develop specific cyctogenetic changes. Despite the strides in etiologic research, this condition remains understudied [6, 7], and established risk factors are few [7]. Though a premenopausal hormonal milieu is needed, what triggers tumor initiation and regulates growth remains a mystery. However, there has been a welcome research push for new, non-invasive treatments [8, 9]. As these become available, it will be important to identify high risk women so that disease can be detected early.

For many conditions, family history is used as a marker of high risk [10]. For example, breast cancer in first-degree relatives is one of the items of the Gail score, a set of questions used to predict breast cancer risk [11]. However, few studies have investigated the predictive value of family history data for fibroids [12-14] (Table 1.). These previous studies reported significant associations between self-reported family history of fibroids and fibroid risk, but none included risk estimates for African Americans, the group in the United States with the highest prevalence of fibroids [7]. African American women are more likely than whites to develop fibroids at a younger age [15], and they have a greater risk of significant symptoms leading to surgery [16, 17]. In addition, they may be less likely than whites to benefit from regression of their disease during the perimenopausal years [18]. Reasons for these ethnic differences are largely unknown, and many believe that genetics may be a contributing factor.

Table 1. Summary of studies evaluating association between family history and fibroids.

Study	Study Design N	Racial Make Up	Fibroid Definition	Family History Definition	Model Adjusted	OR or RR/95%CIa
Templeman et al., 2009(12)	Prospective cohort 80,204 eligible participants N=1790 surgical cases after 11yr follow-up	White-81%b Black-4% Latina -7.4% Asian- 3%.	Pathologic diagnosis	Mother or sister had fibroids Self administrated questionnaire	Race and age	RR:1.42(1.25-1.61)
Sato et al., 2002(13)	Hospital based N=144 surgically diagnosed cases N=288 controls recruited from cancer screening program	100% Japanese	Pathologic diagnosis	First degree relatives had fibroids Self administrated questionnaire	Stratified by parity	< 2 births OR:5.81(2.32-14.57) ≥ 2 births OR:2.11(1.19-3.74)
Van voorhis et al., 2002(14)	Hospital based N=81 Cases N=103 Control recruited women who underwent hysterectomy	95% White	Pathologic diagnosis	Mother had fibroids Self administrated questionnaire	Parity and age	OR:2.85(1.25-6.52)

^aOR:odds ratio, RR:relative risk ratio, 95%CI: 95% confidence interval.

^bRacial breakdown for cohort obtained from Fertil Steril. 2009 Octorber;92(4):1436-1446.

Genetic causes of fibroids are largely unknown, except for fumarate hydratase gene mutations which are seen in rare familial syndromes, but account for only a small fraction of the huge burden of fibroid tumors in women [19]. Twin studies suggest significant heritability of fibroids [20-22], two GWAS (genome wide association studies) implicate possible loci [23, 24], and a linkage study indicates that HMGA-2 is a candidate gene [25], but findings have not been replicated. Thus, though there are certainly heritable contributions to development of common, nonsyndromic fibroids, there are no clinically available markers at this time.

In this manuscript we examine the evidence that self-reported family history of fibroids is associated with presence of fibroids in a large sample of black and white women [26]. This study also offers an opportunity to investigate possible bias in family history studies of fibroids. Biased reporting could arise because a diagnosis of fibroids may lead women to seek out information from family members about their own experiences with fibroids, while women who have never been diagnosed may remain unaware of a family history. The National Institute of Environmental Health Sciences (NIEHS) Uterine Fibroid Study included a questionnaire that asked women about their family history of fibroids, and subsequently screened them for fibroids with standardized ultrasound examinations regardless of any prior diagnosis. This allows us to evaluate the effects of possible differential self-report that could arise depending on knowledge of one's fibroids.

Methods

Study Population

The NIEHS Uterine Fibroid Study was designed to determine the prevalence of fibroids based on ultrasound screening. Study staff enrolled randomly selected women, 35-49 years old, who were members of a prepaid health plan [26]. For this analysis women were excluded if they did not self-identify as African American/black or white, if they did not complete a mail questionnaire which included the family history questions, and if fibroid status was missing. The resulting sample included 1072 women (660 African American, 412 white) (Fig. 1). The research was approved by the National Institute of Environmental Health Sciences and George Washington University Human Subject Review Boards. Participants gave informed consent.

Figure 1.

Population step-down for analysis sample.

Fibroid Status Assessment

To assess fibroid status, premenopausal women were asked to have a transvaginal and transabdominal ultrasound as part of the enrollment clinic visit. Experienced sonographers were trained for the protocol and worked under the direct supervision of a radiologist. Fibroids of ≥ 0.5 centimeter (cm) diameter were identified. Participants who had recently undergone an ultrasound examination at the clinic for clinical purposes were not asked to repeat an examination for the study. For these women, data on presence of fibroids and size of largest were abstracted from the clinic radiology reports. Of the 913 ultrasounds (85% of the total sample of 1072), 198 were recent clinic examinations (18% of the total sample). In addition, 93 women had had a prior uterine surgery, and we abstracted fibroid status data from pathology records for these women (9% of the total sample). For 65 participants who did not complete ultrasound screening examinations and for one woman whose examination was indeterminate, we accepted their self-report of a prior diagnosis (6% of the total sample). Self report of “no fibroids” was not accepted because undiagnosed fibroids were common [26]. Primary statistical analyses were repeated without women whose fibroid status relied on self-report, and results were essentially unchanged.

Family History Assessment

Participants were asked whether either their mother or a sister had fibroids. Those with a positive response for either were considered to have a positive family history.

Statistical Analysis

All analyses were conducted separately for blacks and whites. Descriptive statistics (percentages) were calculated for categorical variables. Because of the high prevalence of fibroids, although odds ratios can be used to evaluate the statistical significance of an association, they over-estimate relative risk [27, 28]. Log-binomial models can be used to provide risk estimates but sometimes these models do not converge [28, 29], as was the case with our data. We used a recommended alternative, the modified Poisson model with robust error variances [30], to estimate the relative risk associated with reported family history. Risk ratios (RR) and 95% confidence intervals (CI) were calculated. Estimates were age-adjusted and also adjusted for other factors previously found to be associated with fibroids in this study population (age at menarche, parity after age 24, body mass index, physical activity) [31]. To investigate how family history estimates might be affected by participant's knowledge of her own fibroid status, we subdivided the cases into two groups depending on whether or not they had had a clinical diagnosis of fibroids prior to reporting the family history information. To investigate how family history associations might be affected by severity of the condition, we limited cases to those with fibroids under 4 cm in largest diameter (numbers were not sufficient to examine women with fibroids ≥4 cm). Within that group of women with relatively homogeneous severity, we again compared associations for the two case groups based on whether or not they had had a clinical diagnosis of fibroids prior to reporting the family history information. Additional sensitivity analyses were conducted to investigate the influence of “opportunity” for a positive family history based on having a mother still living or having at least one sister over age 30 (an age at which fibroids might potentially have been diagnosed). All analyses were conducted using SAS 9.2.1

Results

Of the 1072 women included in this analysis there were 660 African Americans and 412 whites (Table 2). Seventy-seven percent of African American women and 53% of white women had fibroids. Many of the cases were newly identified by study screening (35% of African American cases and 60% of white cases). These cases did not know they had fibroids when they provided family history data. Thirty-eight percent of African Americans and 26% of whites reported a family history of fibroids.

Table 2. Characteristics of participants in the family history analysis of the Uterine Fibroid Study.

Characteristic	African Americans (n=660)		Whites (n=412)
Fibroid status	No.	%	No.	%
No	151	23	193	47
Yes (newly detected fibroids)	177	27	132	32
Yes (self-reported and study confirmed fibroids)	332	50	87	21
Age (years)
35-39	215	32	128	31
40-44	228	35	137	33
45-49	217	33	147	36
Education
≤High school or some college	461	70	49	12
College degree	76	11	65	16
College plus additional	52	8	72	17
Postgraduate	71	11	226	55
Age at menarche (years)
<11	79	12	19	5
11	115	17	62	15
12	178	27	111	27
13	149	23	138	33
14	60	9	48	12
>14	76	12	32	8
Missing	3	--	2	--
Parity
No	139	21	238	58
Yes	521	79	174	42
No. of full-term pregnancies delivered after age 24 years
0	339	51	264	64
1	205	31	58	14
≥2	116	18	90	22
Body mass index
<30	158	24	242	59
30-<35	195	30	96	23
≥35	304	46	74	18
Missing	3	--	--	--
Smoking status
Never	315	47	238	58
Past	156	24	144	35
Current	189	29	30	7
Physical activity
Low (<185)	249	38	107	26
Medium	212	32	144	35
High	101	16	83	20
Very high	94	14	77	19
Missing	4	--	1	--
Family History of Fibroids
None	319	48	211	51
Yes	253	38	108	26
Missing	88	14	93	23

The age adjusted relative risk of fibroids associated with self-reported family history was significantly elevated for both African Americans (RR:1.2, 95%CI 1.1-1.3) and whites (RR:1.3, 95%CI 1.1-1.6). Adjusting estimates further for age at menarche, parity after 24 years of age, body mass index, and physical activity showed little change (Table 3).

Table 3. Associations of self-reported family history of fibroids and fibroids.

	African Americans					Whites
	No. a	RRc	95%CI	aRRd	95%CI	No. a	RRc	95%CI	aRRd	95%CI
Family Historyb
None	226:93	1.00	Ref.	1.0	Ref.	101:110	1.0	Ref.	1.0	Ref.
Yes	211:42	1.2	1.1-1.3	1.2	1.1-1.3	69:39	1.3	1.1-1.6	1.3	1.1-1.5
Missing	72:16	1.2	1.0-1.3	1.2	1.0-1.3	49:44	1.1	0.9-1.4	1.1	0.9-1.4

^anumber of women with fibroids: number of women without fibroids, the multivariable-adjusted analysis included 10 fewer blacks and 3 fewer whites due to missing covariate data.

^bFamily history: mother or sister or both had fibroids.

^cRR, relative risk adjusted for age of woman.

^daRR, adjusted for age, age at menarche, parity after age 24 years, body mass index, physical activity.

We then looked at the family history associations after subdividing the cases according to whether women knew they had fibroids when providing family history information. The adjusted risk ratios for the two groups are shown in Figure 2. There was a strong family history association when cases included those with prior knowledge of their fibroids, but no evidence of a family history association when cases were limited to those who reported family history prior to learning that they had fibroids (cases without prior knowledge). This pattern was seen for both African American and white women.

Figure 2.

The association between self-reported family history of fibroids stratified by prior knowledge of fibroids status among black and white women.

Newly detected fibroids tend to be smaller than previously diagnosed fibroids [26], suggesting less severe disease. Therefore, we considered whether the lack of a family history association among the women with newly detected fibroids might be because they had less severe disease (i.e., family history might only be important in more severe disease). For this analysis we excluded those with the larger fibroids (those with more severe disease), and limited analyses to those with fibroids < 4cm in diameter (a sample of women with less severe disease). Though the sample size is reduced, the same general pattern emerged. The family history association was apparent when cases were limited to those with prior knowledge, but there was no association when cases were limited to newly detected cases (cases that did not know they had fibroids when they reported family history information) (Figure 3).

Figure 3.

The association between self-reported family history of fibroids stratified by prior knowledge of fibroids status for less severe fibroids (< 4 cm diameter) among black and white women.

Finally, we considered the impact of “opportunity” for obtaining family history information (having a living mother and/or a sister ≥30). The results changed little when we adjusted for the opportunity variables (living mother, sister ≥30) or when we excluded the women with little opportunity (7.7% of whites, 5.3% of blacks).

Discussion

A positive family history was significantly associated with a 20% increased risk of having fibroids for African American and a 30% increased risk for white women in our sample. However, knowledge of one's fibroids appeared to influence self-report of family history. The association was attenuated and no longer significant when the analysis was limited to a case group that self-reported family history prior to learning they had fibroids. Severity (inferred by fibroid size) did not appear to account for the observed attenuation of effect. Thus, it is possible that many women learn about their family history of fibroids only after being diagnosed themselves. A diagnosis may spur discussions with family members about the condition, after which the family history is revealed. Women who have not been diagnosed may be unaware of their family history and thus under-report it.

Three previous studies have examined family history of fibroids [12-14] (Table 1). Our results from both black and white full models were similar to those from the prospective cohort study [12]. For comparison, we also calculated odds ratios with logistic regression. Both black (n=660) and white (n=412) women showed an approximate 2-fold elevated odds of fibroids for those self-reporting family history of fibroids compared to those with no family history of fibroids (blacks: OR: 2.07, 95% CI 1.4-3.2; whites: OR: 1.8, 95%CI 1.1-3.0). These results are similar to the prior case-control studies [13, 14] (Table 1). Although all prior studies found significant associations, they could have suffered from the same bias as we seemed to have observed in our data (i.e., non-diagnosed women may be less informed about their family history of fibroids than women with a diagnosis).

Two small studies raise the possibility that more aggressive and detailed assessments of family history data may yield information useful for assessing risk. A Russian study identified and studied family members (mothers, sisters, daughters) of index fibroid cases. The index cases (n=97) that had at least two first degree relatives with fibroids had elevated levels of oncomarkers CA125 and CEA compared to index cases with less than two first degree relatives [32]. A more recent study also looked at clinical and molecular features associated with familial clustering of fibroid tumors. They found that cases with two first degree relatives with fibroids had more severe disease that included elevated VEGF-A expression in tumor tissue compared to cases with fewer than two first degree relatives with fibroids [33].

Our study has both strengths and weaknesses. A unique strength of our study is the ascertainment of fibroids by ultrasound, so there is little misclassification in disease outcome. Another is that we have a sufficiently large sample of African American women that we can report results for that group separately, which has not been done before to our knowledge. The similarity of results for African Americans and whites suggest that our findings may be generalizable to a broad range of women. Limitations include the cross-sectional design and our reliance on self-report for family history information (i.e., it was not verified by other sources). Also our data did not allow us to evaluate severity of fibroids for the participant's mothers or sisters.

In light of the recent and anticipated advances in fibroid treatment [8, 9], better markers of who is likely to develop clinically burdensome tumors and who will remain asymptomatic will become critically important for counseling young women about fibroid-related health issues. Our data suggest that a simple self-report of family history of fibroids is unlikely to be a useful marker of high risk. Other approaches are needed.