Fibroid-Associated Heavy Menstrual Bleeding: Correlation Between Clinical Features, Doppler Ultrasound Assessment of Vasculature, and Tissue Gene Expression Profiles

Abstract

Despite the prevalence of uterine fibroids (Fs), few studies have investigated the links between clinical features and the cellular or molecular mechanisms that drive F growth and development. Such knowledge will ultimately help to differentiate symptomatic from asymptomatic Fs and could result in the development of more effective and individualized treatments. The aim of this study was to investigate the relationship between ultrasound appearance, blood flow, and angiogenic gene expression in F, perifibroid (PM), and distant myometrial (DM) tissues. We hypothesized that angiogenic gene expression would be increased in tissues and participants that showed increased blood flow by Doppler ultrasound. The study was performed using Doppler ultrasound to measure blood flow prior to hysterectomy, with subsequent tissue samples from the F, PM, and DM being investigated for angiogenic gene expression. Overall, PM blood flow (measured as peak systolic velocity [PSV]) was higher than F blood flow, although significant heterogeneity was seen in vascularity and blood flow between different Fs and their surrounding myometrium. We did not find any correlation between PSV and any other clinical or molecular parameter in this study. We identified 19 angiogenesis pathway-related genes with significant differences in expression between F and DM, and 2 genes, matrix metalloproteinase 9 (MMP9) and Neuropilin 2 (NRP2), that were significantly different between F and PM. These results are consistent with subtle differences between PM and DM. Understanding the differences between symptomatic versus asymptomatic Fs may eventually lead to more effective treatments that directly target the source of heavy menstrual bleeding.

Introduction

Uterine leiomyoma are benign neoplasms of the myometrium and are the most common tumors in women of reproductive age. Estimates of the prevalence of leiomyomata, or fibroids (Fs) as they are more commonly called, vary with the population sample and the detection method used. A large US-based population study found a cumulative incidence of more than 80% in black women and almost 70% in white women approaching 50 years of age using diagnostic ultrasound.¹ This incidence concurs with pathological examination of surgical specimens that found that the prevalence was as high as 77%.² It has been estimated that approximately 25% of women with Fs will experience F-related symptoms,³ most commonly heavy menstrual bleeding (HMB), but also pelvic pressure, pain, reproductive dysfunction, and pregnancy complications. Symptomatic uterine Fs pose a major public health, quality-of-life, and economic burden worldwide.⁴

Despite the prevalence and significance of uterine Fs and their variable clinical presentation, few studies have investigated why only a proportion of Fs are symptomatic, or have attempted to correlate clinical symptoms such as HMB with morphology and function of the F-associated vasculature. Very little is known about the cellular or molecular mechanisms that differentiate symptomatic from asymptomatic Fs.

Theories proposed to account for F-associated HMB include reduced myometrial contractility, increased endometrial surface area, and dilated vessels overlying uterine Fs.^5,6 However, there is very little objective evidence to support these proposed mechanisms.^7,8 Uterine Fs are associated with HMB, but there is no consistent relationship between the size and location of Fs and heavy bleeding.⁹ Collectively, this strongly suggests that features other than size or location account for HMB in F uteri.

The presence of intramural Fs in the uterus is likely to significantly alter blood flow within the myometrium and may also alter endometrial blood flow. It has been known for many years that the perifibroid myometrial (PM) region exhibits dilated venules and increased arterial vascularity,¹⁰ which we hypothesize might be linked to increased symptomatic blood loss. Differences in angiogenic gene expression between Fs and normal myometrium have also been demonstrated.¹¹ However, gene expression studies involving the perifibroid region that might identify possible regulatory mechanisms for altered blood flow in this area have not been undertaken.

The aim of this study was to investigate the relationship between blood flow and angiogenic gene expression in F, PM, and distant myometrial (DM) tissues. We hypothesized that angiogenic gene expression would be increased in tissues and participants that showed increased blood flow by Doppler ultrasound. We further hypothesized that F heterogeneity, color Doppler blood flow, and spectral Doppler resistive indices would correlate with differences in gene expression profiles. The study was performed using Doppler ultrasound to measure blood flow prior to hysterectomy, with subsequent tissue samples from the F, PM, and DM being investigated for angiogenic gene expression. A better understanding of F pathophysiology with mechanistic insight into how they cause HMB may potentially lead to novel antiangiogenic treatments for symptomatic uterine Fs.

Methods

Patient Recruitment

Patients with Fs who were scheduled to undergo hysterectomy (n = 6) were recruited via Monash Medical Centre Moorabbin gynaecology outpatients. Ethical approval for the study was obtained from Southern Health Human Research and Ethics Committee B (Project 09272B) and informed consent was obtained from all patients. Clinical details and samples taken from each patient are shown in Table 1.

Table 1.

Clinical Details and Samples Used in Study.^a

ID	Age	Ethnicity	HMB	Pain	Parity	HT	Uterine vol cc	Endometrial Histopathology	Samples Analyzed	Fibroid vol cc
33/10	34	European	Y	N	2	N	181	Mid-secretory	Fibroid 1	234
		descent							Perifibroid 1
									Distant Myo
48/10	46	European	Y	N	4	Y	1184	Inactive	Fibroid 1	135
		descent				Norethisterone			Perifibroid 1
									Fibroid 2	373
									Perifibroid 2
									Distant Myo
73/10	45	Chinese	Y	N	2	N	699	Mid-secretory	Fibroid 1	344
		descent							Perifibroid 1
									Distant Myo
77/10	43	European	N	N	1	N	118	Proliferative	Fibroid 1	23
		descent							Perifibroid 1
									Fibroid 2	28
									Perifibroid 2
									Distant Myo
85/10	52	European	Y	N	2	Y	112	Exogenous hormone effect and residual simple hyperplasia, chronic endometritis	Fibroid 1	11
		descent				Progestagen			Perifibroid 1
									Fibroid 2	3
									Perifibroid 2
									Distant myo
172/10	50	European	Y	Y	2	Y	449	Inactive	Fibroid 2	27
		descent				Ceased tamoxifen 10 weeks prior			Perifibroid 2
									Fibroid 3	20
									Perifibroid 3
									Fibroid 4	146
									Perifibroid 4
									Distant Myo

Abbreviations: HMB, heavy menstrual bleeding (self reported); Y, yes; N, no; HT, hormone treatment within 3 months prior to hysterectomy; uterine vol. cc, uterine volume in cubic centimeters estimated from ultrasound measurement; fibroid vol cc, fibroid volume in cubic centimeters estimated from ultrasound measurement; Distant Myo, myometrial tissue taken as far as possible from any fibroid; PCR, polymerase chain reaction.

^a Clinical details of 6 participants in study. Not all fibroids identified by ultrasound or found at cut up were collected, and not all tissue collected was used for molecular analysis: for example, fibroid 1 from participant 172/10 was not analyzed by PCR-array. Fibroid 1 in participant 33/10 extended beyond the serosal surface of the uterus.

Ultrasound Assessment of Fibroid and Myometrium

Patients recruited to the study underwent B mode, color, and spectral Doppler ultrasound assessment to investigate blood flow in fibroids and PM. Ultrasounds were carried out using C5-2 (transabdominal), C8-4 (endovaginal), and 3D9-v (3D endovaginal) transducers (Philips Electronics, Germany), according to the routine clinical protocol with additional measurements to assess the Fs and surrounding myometrial blood flow. The volume of the uterus in cubic centimeters (cc) was calculated as the volume of an ellipse from 2 coronal and 1 sagittal measurement taken at the maximal diameters of the uterus using the formula 4/3π (R₁·R₂·R₃), where R₁, R₂, and R₃ are diameter/2 (or radius) of the 3 measurements. The number of fibroids was counted and the volume of each fibroid (cc) calculated in the same manner as the volume of the uterus. The location of each fibroid was recorded using the following descriptors: anterior/posterior, right/midline/left, fundal/body/cervix, and minimum distance from endometrium. From this information a schematic diagram was created for each patient documenting the location of the fibroids, so they could be correctly identified at tissue dissection following hysterectomy. A note was made of whether each fibroid distorted the endometrium, with all fibroids that were subsequently selected for molecular analysis falling into this category (Table 2). Using B-mode ultrasound the qualitative characteristics of each fibroid were assessed as follows: the fibroid is predominantly hypoechoic/isoechoic/hyperechoic/mixed compared to the surrounding myometrium. Using spectral Doppler ultrasound the following parameters were recorded in selected vessels: peak systolic velocity (PSV), end-diastolic velocity, and resistance index within each fibroid and in the PM (for 9 of the 11 fibroids—see Table 3). Overall vascularity in the PM region and within the fibroid was further classified as mild, moderate, or marked based on these measurements and on an overall visual assessment.

Table 2.

B-Mode and Doppler Ultrasound Data for the 6 Participants.

ID	Sample	Echotexture of the Fibroid	Echotexture of PF Region	Arterial Flow Within Fibroid	Arterial Flow in PF Region	Flow Within Fibroid Is Therefore:	Fibroid Distorts Endometrium
		Hypoechoic = 1	No Rim = 1	Mild = 1	Mild = 1	> Than Adjacent PF Region = 1
		Isoechoic = 2	Hypoechoic Rim = 2	Moderate = 2	Moderate = 2	= to Adjacent PF Region = 2	Yes = 1
		Hyperechoic = 3	Hyperechoic Rim = 3	Marked = 3	Marked = 3	<Adjacent PF Region = 3	No = 2
		Mixed = 4
33/10	Fibroid 1	2		2		1	1
	Perifibroid 1		2		1
48/10	Fibroid 1	2		2		1	1
	Perifibroid 1		1		2
	Fibroid 2	4		2		1	1
	Perifibroid 2		2		2
73/10	Fibroid 1	2		1		1	1
	Perifibroid 1		2		1
77/10	Fibroid 1	1		1		3	1
	Perifibroid 1		1		2
	Fibroid 2	2		1		3	1
	Perifibroid 2		1		3
85/10	Fibroid 1	1		1		3	1
	Perifibroid 1		1		1
	Fibroid 2	2		1		3	1
	Perifibroid 2		1		2
172/10	Fibroid 2	3		1		3	1
	Perifibroid 2		2		1
	Fibroid 3	2		1		3	1
	Perifibroid 3		1		2
	Fibroid 4	2		1		3	1
	Perifibroid 4		1		2

Abbreviation: PF, perifibroid.

Table 3.

Doppler Ultrasound Data for Fibroid and Perifibroid Regions.^a

ID	Area of Measurement	PSV	EDV	RI
33/10	Fibroid 1	32.2	10.5	0.67
48/10	Fibroid 1	16.8	7.15	0.57
	Fibroid 1	4.5	1.93	0.57
73/10	Fibroid 1	12.9	6.5	0.5
	Fibroid 1	11.3	4.1	0.64
	Fibroid 1	10.8	4.1	0.63
	Perifibroid 1	9.9	4.8	0.52
	Perifibroid 1	25.5	5.8	0.77
	Perifibroid 1	47.6	12.6	0.74
77/10	Fibroid 1	6.3	0.6	0.9
	Fibroid 1	16.8	4.2	0.75
	Fibroid 1	12.2	3.6	0.7
	Perifibroid 1	13.2	3.3	0.75
	Perifibroid 1	36.7	10.5	0.71
	Perifibroid 1	27.1	8.1	0.7
	Fibroid 2	37.6	5.7	0.85
	Fibroid 2	6.6	4.5	0.32
	Fibroid 2	11.4	3	0.74
	Fibroid 2	12.6	5.4	0.57
	Perifibroid 2	48.1	10.5	0.78
	Perifibroid 2	36.1	8.1	0.78
	Perifibroid 2	13.5	4.2	0.69
85/10	Fibroid 1	7.6	2.1	0.72
	Fibroid 1	9.2	2.7	0.7
	Fibroid 1	6.4	0	1
	Perifibroid 1	21.2	4.8	0.77
	Perifibroid 1	29.5	7.5	0.74
	Perifibroid 1	15.7	4.9	0.69
	Fibroid 2	15.6	5.4	0.65
	Fibroid 2	8.9	3.5	0.61
	Fibroid 2	12.6	6.3	0.5
	Perifibroid 2	36.1	13.5	0.63
	Perifibroid 2	32.6	10	0.63
	Perifibroid 2	23.6	6	0.74
172/10	Fibroid 2	8.7	4.2	0.52
	Perifibroid 2	26.5	11.2	0.58
	Perifibroid 2	12.0	6	0.5
	Fibroid 3	4.5	1.8	0.6
	Fibroid 3	4.7	2.1	0.56
	Fibroid 3	9.6	3.2	0.67
	Fibroid 3	12.6	3.2	0.75
	Perifibroid 3	14.9	4.4	0.71
	Fibroid 4	12.0	6	0.5
	Fibroid 4	7.1	3.9	0.45
	Fibroid 4	12.8	7.1	0.45
	Perifibroid 4	45.8	19.3	0.58
	Perifibroid 4	65.1	32.8	0.5
	Perifibroid 4	53.9	22.9	0.57

Abbreviations: PSV, peak systolic velocity; EDV, end diastolic velocity; RI, resistance index.

^a Doppler ultrasound data for 6 participants in study. No perifibroid data were recorded for the first 2 participants in the study.

Sample Collection

Three hysterectomy specimens were from premenopausal women who had not received exogenous hormones for the previous 3 months (patients 33/10, 73/10, 77/10). Of the 3 women who had received exogenous hormones in the 3 months prior to hysterectomy, 2 were premenopausal (patients 85/10, 172/10) and 1 was perimenopausal (patient 48/10). Five women were having hysterectomy for fibroid-related symptoms (HMB) and 1 had an atypical polyp and incidental fibroids (see Table 1 for details).

Immediately following hysterectomy, tissue samples were dissected from 3 areas within each uterus: fibroid F (n = 11), PM (n = 11), and DM (n = 6). Distant myometrium was defined as tissue taken as far as possible from any F. Perifibroid myometrium was defined as directly adjacent to the F. Care was taken to match pathology findings of F size and location with prior ultrasound findings. Samples were collected into RNAlater for transport and storage prior to RNA extraction.

RNA Extraction and Complementary DNA Synthesis

RNA was extracted from tissues using Trizol (Invitrogen, Carlsbad, California) according to the manufacturer’s instructions. RNA solutions were treated with 4 U of Turbo DNase (Life Technologies, Carlsbad, California) for 30 minutes at 37°C, ethanol precipitated, resuspended in RNase-free water quantified using a Nano Drop spectrophotometer. RNA was stored at −80°C until required for use. RNA was converted to complementary DNA (cDNA) using a Transcriptor High-Fidelity cDNA synthesis Kit (Roche, Castle Hill, NSW, Australia) in accordance with the manufacturer’s instructions.

The RT² Profiler PCR Angiogenesis Array, Data Analysis, and Bioinformatics

Complementary DNA samples were used to perform reverse transcriptase–qualitative polymerase chain reaction (RT-qPCR) using a commercially available Human Angiogenesis RT² Profiler PCR array. This custom array covers 84 genes involved in modulating the biological processes related to angiogenesis. The RT-qPCR was run according to the manufacturer’s instructions on an ABI PCR system.

The PCR array data were analyzed using the Bioconductor limma package with fitted linear model, moderated t statistics and Benjamini-Hochberg multiple testing correction. Each of the 11 F and 11 PM samples was first normalized by subtracting the Cts of the corresponding DM samples and then normalized again by subtracting the average results for the control genes Beta-2-microglobulin (B2M); Hypoxanthine phosphoribosyltransferase 1 (HPRT1); Ribosomal protein L13a (RPL13A); Glyceraldehyde-3-phosphate dehydrogenase (GAPDH); and Actin, beta (ACTB). For simplicity, all genes which had missing values were excluded from the analysis. Genes were ranked according to adjusted P values.

Results

Ultrasound Findings

The majority of Fs (10 of the 11 Fs measured) showed equal or increased vascularity in the perifibroid region when compared to vascularity within the F (Table 2 and Figure 1). The majority of flow measured in the F was considered to be mild, while the majority of flow measured in the perifibroid region was considered moderate or marked. Fibroids displayed variable echogenicity by B-mode ultrasound between patients (Table 2 and Figure 2). There was also variability between individual Fs from the same participant.

Figure 1.

Upper panel shows an example of moderate flow in the perifibroid region by Doppler ultrasound, while the lower panel is an example of mild flow within the fibroid. Both images are from participant 172/10, fibroid 4.

Figure 2.

Upper panel (participant 85/10, fibroid 1), middle panel (participant 48/10, fibroid 1), and lower panel (participant 172/10, fibroid 2) demonstrate hypoechoic, isoechoic, and mixed fibroid echogenicity, respectively.

Peak systolic velocity was highly variable when measured in selected vessels in each F and perifibroid region (Table 3). Overall, the mean PSV was significantly higher in the perifibroid region than within the F (30.2 vs 12.0 cm/s, P < .001).

For the bioinformatics analysis of the gene expression data, 3 sets of comparisons were undertaken: relative gene expression in F versus PM samples, F versus DM samples, and PM versus DM samples (Table 4). Of the 84 angiogenesis-related genes on the PCR array, 19 showed statistically significant differences in expression between F and DM (Table 4). Of these, 13 were up in F and 6 were up in DM. A further 13 genes were identified as “of interest” with a P value <.2. Of these, 4 were down in F and 8 were up. There were no genes identified with significant differences in expression between PM and DM, although 4 genes (placental growth factor [PGF], Platelet/endothelial cell adhesion molecule 1 [PECAM1], TIMP1[TIMP metallopeptidase inhibitor 1], and STAB1[Stabilin 1]) were up in PM with a P value of .1 to .2. Two genes were significantly up in F compared to PM, with a further 9 genes (6 up and 3 down), showing differences with a P value of .05 to .2.

Table 4.

Genes Identified by RT-qPCR Using a Commercially Available Human Angiogenesis RT² Profiler PCR Array That Are Differentially Expressed Between Fibroid (F), Perifibroid Myometrium (PM), and Distant Myometrium (DM) After Multiple Testing Correction.

Symbol	F vs DM	P Value	F vs PM	P Value	PM vs DM	P Value	Unigene	Ref seq	Description
CXCL9	Down in F	<.01					Hs.77367	NM_002416	Chemokine (C-X-C motif) ligand 9
KDR	Down in F	<.01					Hs.479756	NM_002253	Kinase insert domain receptor (a type III receptor tyrosine kinase)
S1PR1	Down in F	.01-.05	Down in F	>.05−.1			Hs.154210	NM_001400	Sphingosine-1-phosphate receptor 1
SERPINF1	Down in F	<.01	Down in F	>.05−.1			Hs.532768	NM_002615	Serpin peptidase inhibitor, clade F (alpha-2 antiplasmin, pigment epithelium derived factor), member 1
TEK	Down in F	<.01					Hs.89640	NM_000459	TEK tyrosine kinase, endothelial
THBS1	Down in F	.01−.05					Hs.164226	NM_003246	Thrombospondin 1
EFNA3	Down in F	<.01	Up in F	>.1−.2			Hs.516656	NM_004952	Ephrin-A3
EGF	Up in F	.01−.05	Up in F	>.05−.1			Hs.419815	NM_001963	Epidermal growth factor (beta-urogastrone)
EPHB4	Up in F	.01−.05					Hs.437008	NM_004444	EPH receptor B4
FGF1	Up in F	<.01	Up in F	>.05−.1			Hs.483635	NM_000800	Fibroblast growth factor 1 (acidic)
HAND2	Up in F	.01−.05					Hs.388245	NM_021973	Heart and neural crest derivatives expressed 2
ITGAV	Up in F	<.01					Hs.436873	NM_002210	Integrin, alpha V (vitronectin receptor, alpha polypeptide, antigen CD51)
MMP2	Up in F	<.01	Up in F	>.05−.1			Hs.513617	NM_004530	Matrix metalloproteinase 2 (gelatinase A, 72 kDa gelatinase, 72 kDa type IV collagenase)
MMP9	Up in F	<.01	Up in F	.01−.05			Hs.297413	NM_004994	Matrix metallopeptidase 9 (gelatinase B, 92 kDa gelatinase, 92 kDa type IV collagenase)
NRP2	Up in F	<.01	Up in F	.01−.05			Hs.471200	NM_003872	Neuropilin 2
PDGFA	Up in F	.01−.05					Hs.535898	NM_002607	Platelet-derived growth factor alpha polypeptide
PGF	Up in F	<.01			Up in PM	>.1−.2	Hs.252820	NM_002632	Placental growth factor
TGFB2	Up in F	<.01	Up in F	>.05−.1			Hs.133379	NM_003238	Transforming growth factor, beta 2
THBS2	Up in F	.01−.05					Hs.371147	NM_003247	Thrombospondin 2
CCL2	Down in F	>.1−.2					Hs.303649	NM_002982	Chemokine (C-C motif) ligand 2
COL4A3	Down in F	>.1−.2					Hs.570065	NM_000091	Collagen, type IV, alpha 3 (Goodpasture antigen)
FLT1	Down in F	>.05−.1					Hs.654360	NM_002019	Fms−related tyrosine kinase 1 (vascular endothelial growth factor/vascular permeability factor receptor)
PECAM1			Down in F	>.05−.1	Up in PM	>.1−.2	Hs.514412	NM_000442	Platelet/endothelial cell adhesion molecule
COL18A1	Up in F	>.1−.2					Hs.517356	NM_030582	Collagen, type XVIII, alpha 1
EFNB2	Up in F	>.1−.2					Hs.149239	NM_004093	Ephrin-B2
HIF1A	Up in F	>.1−.2					Hs.597216	NM_001530	Hypoxia inducible factor 1, alpha subunit (basic helix-loop-helix transcription factor)
LAMA5	Up in F	>.1−.2					Hs.473256	NM_005560	Laminin, alpha 5
NOTCH4	Up in F	>.1−.2					Hs.436100	NM_004557	Notch homolog 4 (Drosophila)
PLG			Up in F	>.1−.2			Hs.143436	NM_000301	Plasminogen
TIMP1	Up in F	>.1−.2			Up in PM	>.1−.2	Hs.522632	NM_003254	TIMP metallopeptidase inhibitor 1
TYMP	Up in F	>.1−.2					Hs.592212	NM_001953	Thymidine phosphorylase
STAB1					Up in PM	>.1−.2	Hs.301989	NM_015136	Stabilin 1

Abbreviation: RT−qPCR, reverse transcriptase–polymerase chain reaction.

A cluster dendrogram was created to help visualize degrees of similarity in gene expression between the F and PM samples (Figure 3). This showed that the majority of the PM samples clustered together (7 of 11), although there was a tendency for PM and F samples from the same participant to cluster together in some cases (eg, participants 73/10 and 77/10). We also performed an analysis incorporating the complete data set for all 6 patients (clinical, ultrasound, and gene expression data) but did not identify any other statistically significant correlations.

Figure 3.

Cluster dendrogram showing relative similarity in the overall pattern of gene expression between all fibroid (F) and perifibroid myometrium (PM) samples.

Discussion

This is the first study which has attempted to correlate F-associated ultrasonographic blood flow and other clinical parameters with gene expression. This is also the first study to consider the PM region as a potentially distinct tissue type in the F uterus and to analyze the angiogenic gene expression levels in this area.

The significantly higher mean PSV in the PM region compared to the F demonstrated by this study is consistent with early dye and corrosion casting studies showing a perifibroid “shell” of increased vascularity in this region.^10,12 The PSV is however only 1 component of vascularity, and while every PM region measured had vessels with higher average PSV than the adjacent F, overall differences in vascularity were not always so clear-cut. The observation that blood velocity is typically higher in PM vessels suggests a fundamental difference in the vasculature compared to within the F. We did not find any correlation between PSV and any other clinical or molecular parameter in this study, although numbers are probably too small to permit meaningful conclusions. What these ultrasonographic findings do confirm is that while perifibroid blood flow is generally higher than F blood flow, significant heterogeneity exists in vascularity and blood flow between Fs and their surrounding myometrium. We also observed significant heterogeneity in F and perifibroid echogenicity by B-mode ultrasound. As with the blood flow data, we were unable to identify any significant correlation between echogenicity and any of the other parameters we measured.

This study has identified 19 angiogenesis pathway-related genes with significant differences in expression between F and DM. By comparison only 2 genes, matrix metalloproteinase 9 (MMP9) and Neuropilin 2 (NRP2), were significantly different between F and PM. Thus, despite the fact that direct comparison between DM and PM failed to identify any significant difference in gene expression with the small numbers of participants in our study, our results are consistent with subtle differences between these regions as evidenced by the different gene lists when each of DM and PM were compared to F.

Of the 19 genes that we identified as significantly different between F and DM, most have been linked to F pathophysiology previously, and at least 7 (tyrosine kinase [TEK], thrombospondin 1 [THBS1], ephrin type-B receptor 4 [EPHB4], THBS2, MMP9, MMP2, and PGF) have been reported with the same direction of up- or downregulation between F and myometrium in previous array studies^13–20 This degree of concurrence with the existing literature provides considerable assurance over the validity of our findings. We showed THBS1 was down in Fs compared to myometrium, and THBS2 was up. These results are in broad agreement with earlier studies.^{13,19,21–23} It has been suggested that Fs are caused by an abnormal fibrotic process that is initiated by altered tissue repair in response to injury.^24,25 This is reflected in the aberrant expression of genes that play a role in fibrosis, including thrombospondins, transforming growth factor beta,¹⁶ and extracellular matrix proteins.²² Of the other genes from this previously reported list, MMP2, EPHB4, and PGF have been shown to be upregulated in Fs,^26–28 but there is less convincing evidence for MMP9.

Both sympathetic nerve fibers and the vascular compartment are reduced in Fs compared to myometrium.^29,30 It is therefore not unexpected that genes such as TEK (also known as Tie-2) and KDR (also known as VEGF-R2) are significantly down in F compared to myometrium, since these 2 tyrosine kinase receptors are expressed primarily in endothelial cells. Defects in TEK are associated with inherited venous malformations and the TEK-signalling pathway plays a critical role in endothelial cell–smooth muscle cell communication during venous morphogenesis.³¹ It is thus possible that one or both of these receptors plays a role in the increased vascularity and blood flow seen in the perifibroid region by Doppler ultrasound in the current study. In contrast to TEK and KDR, NRP2 and EFNA3 expression was elevated in Fs compared to myometrium. The neuropilin receptors act as semaphorin coreceptors with plexins to regulate nervous system development as well as receptors for vascular endothelial growth factor family members.³² The ephrins and ephrin-related receptors comprise the largest subfamily of receptor protein tyrosine kinases and have been implicated in mediating developmental events, especially in the nervous system and in erythropoiesis. As for NRP2, elevated EFNA3 in Fs is thus counterintuitive, given that nerves and blood vessels are reduced in this tissue. The EFNA3 receptor family also plays a major role in tumor remodeling,³³ and hence higher levels in Fs may contribute to F growth.

Other genes that we found to have elevated expression in F compared to myometrium were Epidermal growth factor (EGF), Fibroblast growth factor 1 (FGF1), and Platelet-derived growth factor alpha polypeptide (PDGFA). There is evidence implicating EGF as having a central role in mediating the effects of estrogen and progesterone on F growth.^34,35 It has been postulated that both FGF1 and FGF2 play important roles in transformation of normal myometrium into leiomyoma and further growth of this tumor.³⁶ Expression of PDGFA and PDGFB is higher in leiomyoma tissue than in matched myometrial tissue and the expression of PDGF and its receptors in leiomyoma and myometrial tissue varied during the menstrual cycle.³⁷ The same authors speculated that PDGF may play a role in the pathogenesis of leiomyomas through mechanisms that include proliferation and increased expression of extracellular matrix.

Our gene expression results are not in complete concurrence with the literature. While we found integrin, alpha V (ITGAV) to be up in Fs, others have recently reported that this gene and related survival-signalling pathways were downregulated in Fs.³⁸

We identified several genes not previously linked to Fs as differentially expressed compared to DM. The protein from HAND2 is a basic helix-loop-helix transcription factor that plays an essential role in cardiac morphogenesis as well as a role in limb and branchial arch development. More recently a role for HAND2 has been identified in decidualization of human and mouse uterine fibroblast cells.³⁹ The same study identified progesterone as a likely regulator of HAND2 expression during uterine sensitization of the mouse uterus. Elevated expression of a transcription factor in Fs that may be regulated by progesterone and with a role in smooth muscle cell morphogenesis is intriguing and worthy of further study. S1PR1, which was down in Fs, is a receptor for sphingosine-1-phosphate, a bioactive lipid that increases cell survival, proliferation, and angiogenesis. The sphingosine-1-phosphate pathway has established roles in late pregnancy decidua and myometrial contraction,⁴⁰ leading to the possibility that reduced levels of S1PR1 may be linked to failure of normal contraction in F smooth muscle cells. The SERPINF1 (also known as alpha-2 antiplasmin or pigment epithelium–derived factor) is a secreted member of the serpin family that strongly inhibits angiogenesis. In addition, this protein is a neurotrophic factor involved in neuronal differentiation in retinoblastoma cells. Although no role has previously been postulated for SERPINF1 in Fs, there is a recent report of reduced SERPINF1 in peritoneal fluid of women with endometriosis.⁴¹

There are a number of limitations to this study. Assessment of blood flow by ultrasound was problematic with attenuation of signal as the tissue depth increased and wide variability between individual vessels. As a consequence, assessment criteria were usually made as relative comparisons between adjacent tissues rather than absolute values. The limitations of ultrasound are well documented;⁴² however, reproducibility did increase as the study progressed and as the ultrasonographer gained experience with the study protocol. While magnetic resonance imaging is not routinely performed for Fs in a typical clinical setting, it is probable that this imaging modality may be more informative for a similar study in the future.

Most of the women included in this study had multiple Fs and significant anatomical distortion of the uterus and endometrium. It was thus not possible to determine which Fs were responsible for clinical symptoms. This is a common clinical finding and is a significant impediment to linking clinical symptoms with F molecular or cell biology, particularly since Fs often differ in location, size, and appearance within the same uterus. Fibroid heterogeneity was demonstrated with each investigative technique in this study, and while we were unable to produce evidence in the current setting, it seems likely that differences between Fs probably contribute to overall differences in symptomatology.

In this study, participants were recruited from women scheduled for hysterectomy some of whom were taking hormonal medication to reduce symptoms associated with HMB. This medication may have altered gene expression, although the cluster dendrogram analysis (Figure 3) did not show any separation of participants based on hormonal therapy.

Additionally, within the hospital operating lists we were not able to schedule participants at predetermined stages of the menstrual cycle, and with only 6 participants it was not possible to divide into statistically valid cycle stages for subgroup analysis. It is highly likely that with larger numbers and appropriate subgroup analysis for these and other variables, many more genes with significant differences in expression between F, PM, and DM would be identified.

A major limitation of the study was low patient numbers relative to the numbers required for statistically strong bioinformatics analysis of gene array data. In order to better correlate molecular profiles with clinical and ultrasonographic data and to be able to predict features that are associated with HMB, larger groups of both symptomatic and asymptomatic patients are needed. This would be a significant clinical and financial undertaking.

Conclusion

This is the first study that attempts to correlate F-associated clinical features with ultrasound and gene expression data. Despite the small sample size, we were able to show significant differences in angiogenic gene expression between Fs and DM, as well as to a lesser extent PM. While the primary hypothesis that angiogenic gene expression would be increased in tissues and participants that showed increased blood flow by Doppler ultrasound was not proven, we were able to show significant changes in angiogenic gene expression between F and surrounding tissues, with DM and PM having different profiles relative to F. Future studies that correlate clinical symptoms with radiological, cellular, and molecular results are required to allow a better understanding of F heterogeneity, and why some may cause HMB while others do not. The clinical implications of understanding differences between asymptomatic versus symptomatic Fs are significant and may eventually lead to effective treatments that target the source of HMB.