Objective
Reports exist in which cellular leiomyomas (CLs) appear to have clinical characteristics or genetic profiles similar to leiomyosarcomas. This study aimed to determine whether most CLs differ clinically from typical uterine leiomyomas (ULs).
Study Design
A case-control study was conducted with women who underwent surgical procedures between January 1989 and December 2008 and who were diagnosed with a CL (n = 99). Control subjects, who were matched in a 2:1 ratio, were women with a diagnosis of UL (n = 198). Hospital and ambulatory records were reviewed.
Results
In multivariable logistic regression analyses, women with CLs were more likely to have surgery for the indication of enlarging leiomyoma and less likely to have concomitant endometriosis or adenomyosis. Uteri that contained CLs were also more likely to have larger and fewer leiomyomas when compared with control subjects.
Conclusion
CLs have a distinct clinical phenotype compared with ULs and have some characteristics in common with leiomyosarcomas.
Leiomyomas are benign myometrial neoplasms and represent the primary indication for hysterectomy in the United States. They are clinically apparent in 25% of women, although careful pathologic examination of surgical specimens suggests that the prevalence is as high as 77%.
“Typical” or “ordinary” uterine leiomyoma (UL) are composed histologically of smooth muscle cells with bland, uniform, cigar-shaped nuclei that are arranged in interlacing bundles, showing little or no mitotic activity. Leiomyoma variants include mitotically active, cellular, and atypical leiomyomas. Cellular leiomyomas (CLs) represent a subgroup of leiomyoma variants and are defined by the World Health Organization as typical leiomyomas that exhibit hypercellularity. CLs are rare and should not account for >5% of leiomyomas.
It is generally believed that leiomyoma variants have the same benign clinical course as typical ULs. However, in a large case series of uterine leiomyosarcomas, rarely do women with leiomyoma variants appear to have a quasimalignant phenotype and to die of disease over a longer time course than typical leiomyosarcomas, in a manner analogous to borderline ovarian tumors. Additionally, a molecular genetic study defined a group of CLs with deletions of chromosome 1p with transcriptional profiles more like leiomyosarcomas than ULs on hierarchic cluster analysis.
The purpose of the current study was to compare a large series of women with CLs with women with ULs to determine whether CLs have a unique clinical presentation. Furthermore, a data-driven multivariable model of significant findings will allow us to examine features that are associated with CLs in symptomatic women who undergo surgical therapy.
Materials and Methods
This was a retrospective case-control study that was conducted at the Mayo Clinic (Rochester, MN) and was approved by the appropriate institutional review board. All study procedures are in accordance with ethical standards set forth in the revised Declaration of Helsinki.
Using institutional databases, we identified all women who had ULs on final diagnosis after surgery (vaginal or abdominal hysterectomy and abdominal or hysteroscopic myomectomy) for ULs between January 1, 1989, and December 31, 2008, at the Mayo Clinic. Inclusion criteria for the study were authorization of usage of the medical records for research and no presence of malignancy on pathologic examination. The cases with a diagnosis of CL were further identified with the Tissue Registry of the Mayo Clinic and with a pattern search of pathologic reports ( Figure ).
Diagnoses of the identified CL cases were compared manually with the pathologic reports; the sources agreed in 115 of 121 cases (95%). Only pathologically confirmed cases were analyzed. Three CL cases were excluded because of extrapelvic locations of the CLs, and 13 women had a concomitant gynecologic malignancy. Thus, the study group comprised 99 patients with a diagnosis of CL (Figure).
The 12,868 patients with a diagnosis of noncellular UL served as a pool of potential control subjects. Control subjects were matched on the basis of surgeon, surgical procedure, and surgical date (±2 years) in a 2:1 ratio to the CL cases that used an optimal matching algorithm. Matching on the basis of surgeon was done to eliminate confounders of referral patterns and bias of concomitant procedures based on practice style. Records were reviewed to confirm correct coding, and new control subjects were selected to replace ineligible control subjects (false coding of diagnosis or concomitant malignancy). The control group thus comprised a total of 198 patients (Figure).
For each CL case and control subject, we performed a medical record review of both hospital (Rochester Methodist Hospital, Rochester, MN) and ambulatory records (Mayo Clinic, Rochester, MN). Clinical and laboratory values and intraoperative and pathologic information were abstracted. Endometriosis and adenomyosis were recorded on the basis of pathologic diagnosis. Premenopausal status was defined as occurrence of at least 1 menstrual period within 12 months before surgery.
We defined leiomyoma recurrence as either radiographic (magnetic resonance imaging, computed tomography, and ultrasound techniques) or pathologic evidence of new leiomyomas after an initial leiomyoma surgery. Operative notes were reviewed to determine inclusion.
Data were coded and entered into a database (Excel; Microsoft Corporation, Redmond, WA). Statistical analysis was carried out with JMP for Windows (version 7.0.1; Microsoft Corporation) and SAS for Unix (version 9.1.3; SAS Institute, Cary, NC). Means, standard deviations, and medians are reported for continuous variables, and frequency counts and percentages are reported for categoric variables. CL cases were compared with control subjects with the use of the Pearson χ 2 test for nominal or categoric variables and with the use of the 2-sample t test and Wilcoxon rank sum test for normally and nonnormally distributed continuous variables, respectively. Unadjusted odds ratios (ORs) were calculated along with their corresponding 95% confidence interval (CI).
Additionally, we performed a multivariable unconditional stepwise logistic regression analysis for the study sample. All variables with a probability value of < .05 in univariate analysis were considered for building the model; variables with a probability value of < .05 were retained in the final multivariable model. Nominal variables with a fair number of missing information (eg, menometrorrhagia) were modeled with 2 indicator variables. To handle the situation in which a predictor variable was prevalent among the control subjects, but not among the cases (ie, quasi-complete separation of data) and therefore to obtain finite OR estimates, the final multivariable model was fit with the use of a bias reduction method. Adjusted OR estimates were calculated along with their corresponding profile penalized likelihood 95% CI. The c-index, equal to the area under a receiver operating characteristic curve, was used to summarize the overall predictive ability of the final model. All tests were 2-tailed, and a probability value of < .05 was considered statistically significant in all statistical analyses.
Results
Our study sample comprised 297 patients: 99 women with CL and 198 women with UL. Among the CL cases, the surgical procedures consisted of 22 vaginal hysterectomies, 52 total abdominal hysterectomies, 19 abdominal myomectomies, and 6 hysteroscopic myomectomies. The procedures for the 99 CL cases were performed by a total of 17 surgeons. The mean length of follow up for the study sample was 56.8 months (median, 35.0 months). The study sample consisted mainly of white women, who represented 88.9% of the cohort.
Characteristics of the study cohort are presented in Table 1 . Our sample was fairly typical of women with leiomyomas, with most women being in their fifth decade of life, multiparous, and having an increased body mass index. Women with CLs, however, were significantly younger than women with typical leiomyomas (44.7 ± 11.6 vs 48.6 ± 13.0 years; P = .010) and were significantly more likely to report menometrorrhagia and pelvic pressure, compared with women with ULs ( Table 1 ).
| Variable | Cellular leiomyoma (n = 99) | Typical leiomyoma (n = 198) | P value | Odds ratio (95% CI) |
|---|---|---|---|---|
| Age, y a | 44.7 ± 11.6 (45.0) | 48.6 ± 13.0 (46.5) | .010 b | 0.8 (0.6–0.9) c |
| Body mass index, kg/m 2 a | 28.3 ± 7.1 (27.3) | 28.0 ± 6.7 (26.9) | .70 b | 1.0 (0.9–1.2) c |
| Premenopausal, n (%) | 82 (82.8) | 148 (74.7) | .12 d | 1.6 (0.9–3.0) |
| Menorrhagia, n (%) e | 38 (45.8) | 78 (52.7) | .31 d | 0.8 (0.4–1.3) |
| Menometrorrhagia, n (%) e | 17 (20.5) | 15 (10.1) | .030 d | 2.3 (1.1–4.9) |
| Pelvic pressure n (%) | 29 (29.3) | 33 (16.7) | .012 d | 2.1 (1.2–3.7) |
| Surgical indication: enlarging leiomyoma, n (%) | 24 (24.2) | 6 (3.0) | < .001 d | 10.2 (4.0–26.0) |
| Surgery for leiomyoma/leiomyoma-related symptoms, n (%) | 89 (89.9) | 149 (75.3) | .003 d | 2.9 (1.4–6.1) |
| Concomitant adenomyosis, n (%) | 0 | 23 (11.6) | < .001 d | 0.04 (0.002–0.6) |
| Concomitant endometriosis, n (%) | 8 (8.1) | 46 (23.2) | .001 d | 0.3 (0.1–0.6) |
| 1 leiomyoma f | 52 (53.6) | 70 (35.9) | .004 d | 2.1 (1.3–3.4) |
| Diameter largest leiomyoma, cm a | 8.8 ± 4.8 (8.0) | 4.7 ± 4.5 (3.4) | < .001 g | 2.3 (1.8–3.0) c |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
