https://orcid.org/0000-0002-3218-1075
Uterine leiomyomas (UL), also known as fibroids or myomas, are benign smooth muscle tumors of the uterine wall. Despite their benign nature, the effects of UL on quality of life may be profound including heavy uterine bleeding, abdominal discomfort and even infertility. Detected in ~70% of women of reproductive age with an estimated annual cost of 5.9–34.4 billion dollars in the USA, UL represent a major public healthcare problem (1,2). Definitive treatment for UL is surgery, however, medical management including the use of GnRH agonists and selective progesterone receptor modulator ulipristal acetate (UPA) can be considered to reduce symptoms prior to surgery or for non-surgical candidates (3,4). Given multiple UL from the same individual may have variable response to medical therapy, the molecular subgroup of UL might be contributory to each tumor’s unique response to therapy.
UL are genetically heterogenous group of tumors with distinct molecular subgroups harboring MED12 (70%), HMGA2 (10%), FH (1–2%) and SRCAP complex gene (2%) alterations (5–7). Their molecular features are shown to have some clinicopathologic correlations such as the larger, rapidly growing and highly vascular nature of HMGA2-overexpressing UL and the smaller size and subserosal location seen in MED12-mutated UL (8–10). To analyze the association of molecular UL subgroups with treatment response, Kolterud et al. (11) studied 101 UL (from 81 women) surgically removed after UPA treatment.
The authors divided the tumors into three main molecular categories: MED12-(51%), HMGA2-(22%) and non-HMGA2/MED12-UL (non-MH, 27%). The treatment response was determined based on the changes in the tumor size obtained from imaging and pathology reports, as well as written assessments by the clinicians blinded to molecular groups. Overall, MED12-UL were 4.4 times more likely to shrink than HMGA2-UL (95% confidence interval 1.37–13.9; P = 0.013) regardless of their initial size, total number of tumors or morphological characteristics (necrosis, edema or hemorrhage). In addition, the age and parity of women, and patient cohort (Sweden or Finland) did not show any association.
To further understand the difference in the treatment response between MED12 and HMGA2-UL, Kolterud et al. analyzed progesterone pathway gene expression and methylation, given that UPA is a progesterone receptor modulator. MED12-UL showed increased expression of progesterone receptor (PGR), as well as hypomethylation of PGR binding sites in the tumor genome, suggestive of active PGR signaling (Fig. 1). In addition, MED12-and HMGA2-UL showed distinct progesterone pathway gene expression patterns by principal component analysis. Combined together, multiple layers of data suggest a difference in PGR activity between the two molecular subgroups, which may explain their distinct treatment response to UPA. Of note, in parallel to the authors’ findings, an earlier study suggests MED12 mutation may result in increased interaction with PGR (12).
Schematic diagram of ulipristal acetate (UPA) treated HMGA2- and MED12-UL showing increased progesterone receptor (PGR) expression with hypomethylation of PGR binding sites in the setting of more tumor shrinkage in MED12-UL compared with HMGA2-UL.
Although the premise of the study by Kolterud et al. is very exciting, there are some limitations. In this retrospective and observational study, the changes in tumor size before and after therapy were not systematically recorded by the same methodology and clinician (e.g. comparison of an imaging size to gross pathologic examination or different operators recording size for each imaging study are not ideal especially for subtle changes) and sometimes subjective notes by the clinicians such as ‘excellent tumor shrinkage’ without actual measurements were used. In addition, the study population is probably skewed with less responsive tumors because only cases with surgical removal were analyzed. One could argue that the tumors with best therapy response might not have required a subsequent surgery and therefore not included in this study. Furthermore, even though the tumor responses did not change between patient cohorts (Sweden or Finland) perhaps the significance of these findings based on ethnicity still needs to be explored, especially given that black patients have a significantly increased risk for developing UL than white women and present with more severe symptoms (2). In addition, the rare molecular subgroups in the non-MH category were not studied extensively due to limited sample size, and the location of the tumors were not taken into account. There are also practical challenges for real-world application of these results, such as the feasibility of detecting the molecular subgroup of UL prior to surgery, which can eventually be resolved by cell free tumor DNA testing or hysteroscopic sampling of the tumor if it is submucosal. However, this can be complicated in the setting of multiple UL in the same patient. Lastly, this study only analyzes the UPA as the medical therapy and the use of this drug has recently been restricted by the European Medicines Agency due to rare side effect of liver injury. Other progesterone receptor modulators for UL treatment are currently investigated and further studies are warranted for other medications.
This is the first study showing evidence that UL genomic subclasses may predict medical treatment response. Despite its retrospective and somewhat subjective nature, this report represents the first fundamental and much-needed step towards future prospective studies that will be evolving in the setting of rapidly advancing molecular pathology and drug development/delivery fields, with the ultimate goal of eventually bringing the management of UL, a major public health problem, into the precision medicine era.
Conflict of interest statement: None declared.
