Objective
The objective of the study was to determine the contribution of submucosal fibroids (SMs) to heavy menstrual bleeding (HMB) and anemia among women with HMB.
Study Design
Our retrospective study included premenopausal women who presented to a tertiary care center for HMB between January 2007 and October 2011. All women in this cohort underwent flexible office hysteroscopy (n = 1665) and 259 (15.6%) had SMs. We also reviewed the clinical ultrasounds (n = 914) from these women to determine whether SMs (n = 148) or any fibroids (n = 434) were present in the uterus. Clinical evaluation of bleeding included hemoglobin and pictorial blood loss assessment charts.
Results
In our cohort, hysteroscopically diagnosed SMs were associated with significantly lower hemoglobin (adjusted difference −0.35 g/dL; 95% confidence interval [CI], −0.56 g/dL to −0.13g/dL) and higher risk of anemia (odds ratio [OR], 1.46; 95% CI, 1.04−2.03). Women with ultrasound-diagnosed SMs had lower hemoglobin and anemia, but results were not significant once adjusted for confounders (hemoglobin: adjusted difference −0.21 g/dL; 95% CI, −0.47g/dL to 0.06 g/dL; and anemia: OR, 1.28; 95% CI, 0.82−1.97). Ultrasound-diagnosed fibroids anywhere in the uterus were not associated with hemoglobin ( P = .7) or anemia ( P = .8). Self-reported pictorial blood loss assessment charts scores did not differ between women with and without fibroids diagnosed by either hysteroscopy or ultrasound ( P = .4 and P = .9, respectively).
Conclusion
SMs were related to lower hemoglobin and higher risk of anemia but not self-reported bleeding scores. Diagnostic modality was important: hysteroscopically diagnosed SMs had lower hemoglobin and more anemia than ultrasound-diagnosed SMs. This may explain the inconsistent results in the literature.
Submucosal fibroids (SMs) are presumed to cause heavy menstrual bleeding (HMB) leading to anemia, but not all research supports this. The definition of SM can vary from distorting the uterine cavity to being adjacent to it, and the classification depends on the diagnostic modality used. Hysteroscopy and sonohysterography are able to distinguish distortion of the uterine cavity better than ultrasound alone. Thus, studies that use hysteroscopy may define SMs differently than those that use ultrasound.
Clinical studies indicate that the removal of SMs that distort the uterine cavity will decrease HMB. Bleeding was controlled by myomectomy in 81% of women in one study and 94% in a second study. Studies of long-term outcomes of hysteroscopic myomectomy demonstrate a 14-86% improvement rate in HMB inversely related to fibroid size. However, other studies have not found an association between HMB and submucosal location. An ultrasound-based cross-sectional study concluded that size of the fibroid, but not submucosal classification, was related to gushing-type bleeding. A sonohysterography-based study found that both intramural fibroids and SMs were fairly equally related to bleeding.
The Mayo Clinic hysteroscopy database is a unique resource well suited to determining the association of HMB and anemia caused by SMs. We hypothesized that hysteroscopically diagnosed SMs would contribute to lower hemoglobin, higher risk of anemia, and higher pictorial blood loss assessment chart (PBLAC) scores among women with HMB. We also hypothesized that ultrasound-diagnosed fibroids would have less impact on bleeding than hysteroscopically diagnosed tumors and would depend on fibroid location. We have more than 3500 patients recorded who have undergone a hysteroscopic evaluation for abnormal uterine bleeding over the past 5 years. On these same women, we have systematically recorded the available clinical ultrasound data to compare with data from hysteroscopy.
Materials and Methods
We collected clinical data on women who were treated by our physician group between January 2007 and October 2011. All women had undergone a flexible office hysteroscopy for evaluation of HMB ( Figure ). The institutional review board approved the protocol and only women who provided research authorization were included in the study. Demographic, clinical, and laboratory variables were abstracted from the patient electronic medical records. During their clinical visit, participants were routinely asked about the use of iron supplements and hormones and their risks of endometrial hyperplasia including obesity, infertility, diabetes, and hypertension. A standardized evaluation form was implemented early in 2009 to document these results.
Office hysteroscopy was performed using flexible Olympus Hyf-XP (Tokyo, Japan) 3.1 mm scopes without local anesthesia (unless necessary or requested by patient) and using saline as the distending media. Saline was gravity fed with intracavitary pressures of 50 mm Hg or less. Hysteroscopic diagnoses were categorized on visualization of cavity using 18 predetermined diagnoses including submucosal fibroid. Inadequate hysteroscopies in which the visualization of the cavity did not occur were documented (n = 10) and were excluded from analyses.
The cohort was divided into women with any documented SMs by hysteroscopy and those without any SMs. The class of SM described by the surgeon was recorded using the European Society for Gynecologic Endoscopists’ classification system, with class 0 being completely in the cavity, class 1 more than 50% in the cavity, and class 2 less than 50% in the cavity.
A PBLAC was used to document bleeding at the clinical visit. This is a validated measure of blood loss that asks women to quantify the number of tampons and pads used per day for each day in their cycle. The number of clots and menstrual accidents is also quantified and the weighted scores added together. A score of more than 100 corresponds to 80 mL blood loss and represents HMB. Hemoglobin and iron store measures, including mean corpuscular volume and ferritin, were recorded when available from routine laboratory draws done within 1 year prior to or at the time of hysteroscopy. We performed a sensitivity analysis comparing PBLAC scores to anemia and hemoglobin to evaluate the relationship between the subjective and objective measures of bleeding.
Additionally, we collected data from patients in the hysteroscopy cohort who underwent ultrasound for clinical purposes within the past year. Standard protocol for ultrasound is to view and measure the fibroids in 2 views and document volume by cine clip. Type of fibroids (intramural, subserosal, and submucosal) was recorded from ultrasound reports. For missing data, images were reviewed and data recorded by 1 of 2 gynecologists (S.K.L., K.P.). Fibroids on ultrasound report that were listed as abutting, distorting, or in the uterine cavity were classified as submucosal fibroids for this study. Thirty-six fibroids were listed as unknown and were reviewed by one of the authors (S.K.L.). Three fibroids could not be classified and the other 33 were categorized based on image review.
The first comparison was made between women with ultrasound-diagnosed submucosal fibroids vs women with nonsubmucosal fibroids or no fibroids limited only to women with an ultrasound (to limit misclassification). Second, we compared women with any ultrasound-diagnosed fibroids (submucosal, intramural, and subserosal) with those without fibroids limited only to women who had an ultrasound done.
Lastly, because all women with an ultrasound also had a hysteroscopy, we created a concordance variable that included women who had SMs on both ultrasound and hysteroscopy. This concordance variable was then used to run a sensitivity analysis on the outcomes to determine whether SM fibroids on ultrasound, confirmed by hysteroscopy, would lead to stronger bleeding outcomes.
Statistical analysis
Categorical and continuous data were analyzed using χ 2 and Student t tests, respectively. Distributions of the continuous data were evaluated. An SD is the method of dispersion reported for mean values of continuous data. PBLAC scores were skewed; thus, we used log transformation to compare groups and created categories to display data. Categories were based on quantile.
Univariate and multivariate analyses were performed for the outcomes of hemoglobin, anemia (defined as hemoglobin <12 g/dL), and PBLAC score greater than 500. Step-wise backward regression models were used to determine significant covariates for each set of analyses. For the continuous outcome of hemoglobin, a generalized linear model (Gaussian link) was used such that the outcome is a reduction in hemoglobin (grams per deciliter). For the binary outcomes of anemia and PBLAC score, a logistic regression model was performed. Significance level was set at a value of P < .05. For the 259 women with hysteroscopically diagnosed fibroids, we had an 88% power to detect a difference of 0.5 g/dL hemoglobin using a 2-sided Student t test with a type I error rate of 5%. The power was greater than 90% for the same testing for women with ultrasound-diagnosed SM.
Results
More than 3500 women had an office hysteroscopy between Jan. 1, 2007, and Oct. 21, 2011. More than 98% of the hysteroscopies were performed by 1 of 5 physicians following a standard protocol. The 1746 women who had a hysteroscopy performed for the indication of HMB were included in this cohort. Women who did not provide research authorization (n = 71) were excluded, and 10 hysteroscopies were inadequate, leaving 1665 for analysis ( Figure ). The mean age of women undergoing hysteroscopy was 43.18 ± 6.85, mean body mass index 29.14 ± 7.87, and most were multigravid and multiparous ( Table 1 ). Participants were able to have indications for hysteroscopy in addition to HMB; irregular menstrual bleeding (IMB) was the most common (n = 390). Nearly 70% reported pain during menses; however, data on painful menses were missing for 30% of the women.
| Factor | No SM | SM | Total | P value |
|---|---|---|---|---|
| n (%) | 1406 (84.4) | 259 (15.6) | 1665 | |
| Age (mean, SD) | 42.83 (6.98) | 45.07 (5.76) | 43.18 (6.85) | < .001 |
| Gravida | ||||
| 0 | 166 (11.86) | 41 (15.89) | 207 (12.48) | .08 |
| 1 | 127 (9.07) | 29 (11.24) | 156 (9.41) | |
| ≥2 | 1107 (79.07) | 188 (72.87) | 1295 (78.11) | |
| Missing | 6 | 1 | 7 | |
| Para | ||||
| 0 | 205 (14.72) | 58 (22.57) | 263 (15.94) | < .001 |
| 1 | 130 (9.33) | 34 (13.23) | 164 (9.94) | |
| ≥2 | 1058 (75.95) | 165 (64.20) | 1223 (74.12) | |
| Missing | 13 | 2 | 15 | |
| BMI (mean, SD) | 29.28 (7.92) | 28.39 (7.56) | 29.14 (7.87) | .1 |
| Hormone use | ||||
| No | 1164 (84.65) | 185 (72.83) | 1349 (82.81) | < .001 |
| Yes | 211 (15.35) | 69 (27.17) | 280 (17.19) | |
| Missing | 31 | 5 | 36 | |
| Iron use | ||||
| No | 1133 (80.81) | 182 (70.27) | 1315 (79.17) | < .001 |
| Yes | 269 (19.19) | 77 (29.73) | 346 (20.83) | |
| Missing | 4 | 0 | 4 | |
| Metrorrhagia | ||||
| No | 1054 (74.96) | 221 (85.33) | 1275 (76.58) | < .001 |
| Yes | 352 (25.04) | 38 (14.67) | 390 (23.42) | |
| Dysmenorrhea | ||||
| No | 310 (31.16) | 48 (30.00) | 358 (31.00) | .8 |
| Yes | 685 (68.84) | 112 (70.00) | 797 (69.00) | |
| Missing | 411 | 99 | 510 | |
| Bleeding parameters | ||||
| Hemoglobin (mean, SD) | 12.62 (1.48) | 12.12 (1.70) | 12.54 (1.52) | < .001 |
| Anemia | ||||
| No | 888 (73.09) | 137 (61.71) | 1,025 (71.33) | .001 |
| Yes | 327 (26.91) | 85 (38.29) | 412 (28.67) | |
| Missing | 191 | 37 | 228 | |
| PBLAC ≥500 | ||||
| No | 680 (61.43) | 111 (58.73) | 791 (61.03) | .5 |
| Yes | 427 (38.57) | 78 (41.27) | 505 (38.97) | |
| Missing | 299 | 70 | 369 | |
Submucosal fibroids diagnosed by hysteroscopy
In this cohort, 259 women (15.6%) were diagnosed with SM by hysteroscopy. Sixty-three were class 0, 110 were class 1, 52 were class 2, and 25 were not documented by the physician. Women with SMs were significantly older and were more likely to be nulliparous ( Table 1 ). They were more likely to be using hormones or iron therapy. IMB was significantly more common for women without SMs, but painful menses did not differ between groups.
Women with SMs had lower hemoglobin (difference in mean, 0.5 g/dL; P < .001; Table 1 ) and were significantly more likely to have anemia (38.29% vs 26.91%; P = .001). Multiple measures of iron deficiency including ferritin and mean corpuscular volume were consistent with these findings. In multivariate analyses ( Table 2 ), SMs were associated with a lower hemoglobin (adjusted difference, −0.35; 95% confidence interval [CI], −0.56 to −0.13; P = .002) and an increased odds ratio for anemia (adjusted odds ratio [OR], 1.46; 95% CI, 1.04−1.97; P = .03). Both hemoglobin and anemia showed trends by the class of SM ( Table 3 ).
| Bleeding parameter | Unadjusted difference, (95% CI) | Adjusted difference, (95% CI) a | Adjusted difference, (95% CI) b | P value (adjusted) b |
|---|---|---|---|---|
| Hemoglobin | ||||
| SM by hysteroscopy c | −0.49 (−0.71 to −0.28) | −0.45 (−0.68 to −0.23) | −0.35 (−0.56 to −0.13) | .002 |
| SM by US c | −0.31 (−0.58 to −0.05) | −0.28 (−0.56 to −0.01) | −0.21 (−0.47 to 0.06) | .1 |
| Any fibroids by US c | −0.05 (−0.25 to 0.14) | 0.01 (−0.20 to 0.22) | 0.04 (−0.16 to 0.25) | .7 |
| Unadjusted OR, (95% CI) | Adjusted OR, (95% CI) a | Adjusted OR, (95% CI) b | P value (adjusted) b | |
| Anemia | ||||
| SM by hysteroscopy d | 1.68 (1.25−2.27) | 1.62 (1.18−2.24) | 1.46 (1.04−2.03) | .03 |
| SM by US c | 1.50 (1.00−2.24) | 1.38 (0.90−2.10) | 1.28 (0.82−1.97) | .3 |
| Any fibroids by US c | 1.15 (0.84−1.57) | 1.01 (0.72−1.42) | 0.96 (0.68−1.37) | .8 |
| PBLAC >500 | ||||
| SM by hysteroscopy e | 1.12 (0.82−1.53) | 1.16 (0.84−1.61) | 1.15 (0.83−1.59) | .4 |
| SM by US f | 1.10 (0.72−1.67) | 1.05 (0.68−1.62) | 1.03 (0.67−1.59) | .9 |
| Any fibroids by US e | 1.16 (0.85−1.58) | 1.12 (0.80−1.56) | 1.10 (0.79−1.53) | .6 |
a Adjusted models not including iron use
b Adjusted models including iron use
c Adjusted for age, parity, hormone use, and IMB
d Adjusted for age, parity, hormone use, IMB, and body mass index
e Adjusted for parity, IMB, and body mass index
| Variable | Class 0 | Class 1 | Class 2 | Non-SM fibroid |
|---|---|---|---|---|
| Hemoglobin a | 11.99 (1.75) | 12.11 (1.87) | 12.41 (1.32) | 12.62 (1.47) |
| Anemia a | ||||
| No | 30 (56.60) | 59 (61.46) | 32 (72.73) | 900 (73.05) |
| Yes | 23 (43.40) | 37 (38.54) | 12 (27.27) | 332 (26.95) |
| Missing | 10 | 14 | 8 | 192 |
| PBLAC ≥500 | ||||
| No | 28 (63.64) | 52 (60.47) | 16 (43.24) | 690 (61.72) |
| Yes | 16 (36.36) | 34 (39.53) | 21 (56.76) | 428 (38.28) |
| Missing |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
