Adjuvant Treatment Recommendations in Early-Stage Endometrial Cancer: What Changes With the Introduction of The Integrated Molecular-Based Risk Assessment

Adjuvant therapy recommendations for endometrial cancer were historically based on the individual patient’s risk of disease recurrence using clinicopathologic factors such as age, stage, histologic subtype, tumor grade, and lymphovascular space invasion. Despite the excellent prognosis for early stages, considerable under- and overtreatment remains. Integrated genomic characterization by the Cancer Genome Atlas (TCGA) in 2013 defined four distinct endometrial cancer subgroups (POLE mutated, microsatellite instability, low copy number, and high copy number) with possible prognostic value. The validation of surrogate markers (p53, Mismatch repair deficiency, and POLE) to determine these subgroups and the addition of other molecular prognosticators (CTNNB1, L1CAM) resulted in a practical and clinically useful molecular classification tool. The incorporation of such molecular alterations into established clinicopathologic risk factors resulted in a refined, improved risk assessment. Thus, the ESGO/ESTRO/ESP consensus in 2020 defined for the first time different prognostic risk groups integrating molecular markers. Finally, the feasibility and clinical utility of molecular profiling for tailoring adjuvant therapy in the high-intermediate-risk group is currently under investigation (NCT03469674).

outcomes, and type II tumors (30-40%), more common in nonobese women, associated with an atrophic endometrium, with aggressive clinical behavior and poor outcome (2).
Adjuvant therapy recommendations have traditionally been based on the individual patient's risk of disease recurrence using clinicopathologic factors such as age, stage, histologic subtype, tumor grade, and lymphovascular space invasion (LVSI) (3,4). In particular, the ESMO/ESGO/ESTRO (European Society for Medical Oncology-European Society for Radiotherapy and Oncology-European Society of Gynaecological Oncology) consensus in 2016 proposed five risk groups to guide adjuvant therapy use (low, intermediate, high-intermediate, high, advanced/metastatic) (4).
However, a small but substantial number of patients with favorable prognostic background unexpectedly experience recurrence of disease and poor survival (5)(6)(7)(8), and it has been calculated that up to 10% of them will experience distant metastasis (7).
On the other hand, a not-negligible number of patients with unfavorable prognostic factors that are usually treated will never experience recurrence: in particular, seven high-intermediaterisk patients need to undergo vaginal brachytherapy (EBRT) to prevent one recurrence (7).
In 2013, the Cancer Genome Atlas (TCGA) defined four distinct EC subgroups (POLE mutated, microsatellite instability, low copy number, and high copy number) with possible prognostic value, and many others confirmed these data in external cohorts (3-5, 7, 8). Molecular risk classes are not completely superimposable with clinicopathological categories, but the combination of both models has been shown to perform better in terms of prognosis than the single ones by themselves (7).
The most recent ESGO/ESTRO and the European Society of Pathology (ESP) ESP recommendations, published at the end of 2020, incorporated molecular and clinicopathological features in an integrated classification system in order to guide adjuvant treatment choices (9).

A COMPREHENSIVE GENOMIC AND TRANSCRIPTOMIC ANALYSIS THROUGH NEXT-GENERATION SEQUENCING (NGS): A GENOMIC CLASSIFICATION
In 2013, the Cancer Genome Atlas Research Network (TCGA) has reported a comprehensive genomic and transcriptomic analysis of 373 EC cases, mainly endometrioid (307) (8).
This characterization categorized EC tumors into four genomic classes with different molecular and prognostic profiles. Such molecular analyses were proven feasible (>96%) and highly reproducible in external cohorts of patients (7,10,11). The distribution of histologic subtypes into the four molecular classes is reported in Figure 1.

POLEmut
This molecular class is defined by pathogenic mutations in the exonuclease domain of DNA polymerase epsilon (10).
This gene encodes a catalytic subunit of DNA polymerase epsilon involved in nuclear DNA replication and repair. The most common alterations in POLE detected in EC samples are hotspot mutations at P286R, V411L, S297F, A456P, and S459F (10,(12)(13)(14).
Overall, this genomic class is associated with excellent prognosis. It accounts for less than 10% of all EC cases, and it is associated with low copy-number aberrations and a very high mutational burden (232 × 10 −6 mutations/Mb).
The associated morphological characteristics of this subgroup include high rate of tumor-infiltrating lymphocytes and/or peritumoral lymphocytes, morphologic heterogeneity/ ambiguity, bizarre/giant tumor cell nuclei, endometrioid histotype but also clear cell carcinomas, undifferentiated carcinomas, and carcinosarcomas (13,15).
Approximately 65% of this molecular class is associated with intermediate and high-risk phenotype according to ESMO 2013 classification (16). In particular, it is frequently associated with grade 3 endometrioid cancers. Given the favorable prognosis of this subgroup, no adjuvant treatment could be suggested, reducing the possible overtreatment, particularly in the highintermediate-and high-risk group. The PORTEC-4a trial (ISRCTN11659025) will answer whether this strategy is safe and efficient in the high-intermediate-risk subgroup (17).

Microsatellite Unstable (MSI Hypermutated)
This molecular class is characterized by the presence of microsatellite instability.
MSI represents the phenotypic evidence that DNA mismatch repair (MMR) is not functioning normally. The MMR is a system for recognizing and repairing erroneous insertion, deletion, and misincorporation of bases that can arise during DNA replication and recombination, as well as repairing other forms of DNA damage.
The most common alteration in MMR detected in EC samples is MLH1 promoter methylation (8).
O v e r a l l , t h i s g e n o m i c c l a s s i s a s s o c i a te d w i t h intermediate prognosis.
It encounters for 25-30% of all EC cases and is characterized by low copy-number aberrations and high mutational burden (18 × 10 −6 mutations/Mb).
Approximately 30% of this molecular class is associated with the low-risk phenotype according to ESMO 2013 classification (16).

Copy-Number High (Serous-Like)
This genomic class is defined mainly by TP53 mutations (8).
P53 gene encodes for p53 protein (TP53) mainly involved in cell cycle arrest, DNA repair, senescence, apoptosis induction, but also cell metabolism regulation and cell response to oxidative stress (23).
Overall, this genomic class is associated with unfavorable prognosis. It encounters for 25-30% of all EC cases and is characterized by a low mutation rate (2·3 × 10 −6 mutations/Mb) and high copy-number aberrations.
The associated morphological characteristics of this class include serous, endometrioid, and mixed serous and endometrioid histology, grade 3, poor inflammatory stroma (23).
Approximately 25% of this molecular class is associated with low-and intermediate-risk phenotypes according to ESMO 2013 classification (16). About 4% of EC are classified as multiple classifier at molecular profiling, and when POLE and p53abn coexist, the prognosis is driven by POLE. In the same way, when MMR-d and p53abn coexist, the prognosis is driven by MMRd (24).

Copy-Number Low
This genomic class comprises mainly microsatellite-stable cancers characterized by frequent CTNNB1 mutations (8).
O v e r a l l , t h i s g e n o m ic c l a s s i s a s s o c i a te d w i t h intermediate prognosis.
It encounters for 30-40% of all EC cases and is characterized by a low mutational burden (2.9 × 10 −6 mutations/Mb) and low copy-number aberrations.
Approximately 50% of this molecular class is associated with the low-risk phenotype according to ESMO 2013 classification (16).

FROM NGS TO IMMUNOHISTOCHEMISTRY (IHC): THE MOLECULAR EC CLASSIFICATION
Methodologies used for the TCGA study are costly, complex, and unsuitable for wider implementation in clinical practice. In 2015, a pragmatic molecular classifier based on surrogate immunohistochemistry assays was developed and validated in internal and external cohorts (10,11,26,27). It was aimed at replicating and replacing the TCGA classification, which relied on whole-exome sequencing (WES). This approach was tested by multiple study groups, which makes evidences concerning its feasibility and reliability especially robust (28).
The "high copy number" group was identified as p53abnormal (p53abn) determining p53 status by IHC testing; the subgroup obtained, however, was not completely equivalent to the TCGA one.
No surrogate was found for POLEmut detection; thus, NGS was maintained.
The "low copy number" group was determined by exclusion and called non-specific molecular profile (NSMP).

INTEGRATED CLINICOPATHOLOGIC AND MOLECULAR CLASSIFICATION: THE ESGO/ESTRO/ESP 2020 RISK CLASSIFICATION AND ADJUVANT TREATMENT RECOMMENDATIONS
The integration of molecular and clinicopathological factors in early-stage ECs in various PORTEC trials cohorts resulted in a stronger model with improved risk prognostication (7).
In particular, the AUC of the integrated molecular risk assessment showed a substantial improvement in predicting locoregional recurrence, distant recurrence, and overall survival compared to clinicopathological classification alone.
Its main implication is to guide clinicians' choices in terms of fertility-sparing treatments, surgery, adjuvant therapy, and surveillance in order to improve outcomes for women with EC.
In the light of available evidences, the ESGO/ESTRO/ESP decided to jointly update EC management evidence-based guidelines, implementing the use of molecular classification.
Risk group classification includes both cases that undergo molecular profiling and cases who did not. If molecular classification tools are not available, traditional pathologic features are used to classify EC patients. The main characteristics of the large trials included in the consensus, which guided treatment decision making, are summarized in Table 1.
Clinicopathological factors include the following: A consensus definition for LVSI in the literature is lacking. It reported good inter-observer agreement in discriminating "true LVSI" from "LVSI mimics" and in grading the extent of LVSI through a semiquantitative system (33). Nevertheless, some problematic cases exist. In addition, substantial LVSI in EC seems to have a stronger prognostic significance than focal LVSI (34)(35)(36).
Overall, the new ESTRO/ESGO/ESP guidelines published in 2020 integrate molecular into clinical classification and encouraged molecular classification in all EC especially highgrade tumors with only POLE mutation analysis possibly omitted in low-risk and intermediate-risk carcinoma with lowgrade histology. Based on this, p53abn tumors with myometrial invasion are considered and treated as high-risk patients with chemotherapy or the combination of chemotherapy and radiotherapy. Stage I-II POLEmut ECs without residual disease are considered low-risk patients for which no adjuvant treatment is recommended. Routine lymphadenectomy for nodal staging purposes is generally not recommended for this group (37,38). Sentinel lymph node biopsy can be considered for staging purposes, but it can be omitted in cases without myometrial invasion (38,39). The incidence of recurrence after surgery alone is <5% (40). No adjuvant treatment is recommended for this group.

Intermediate-Risk Class
This risk class includes patients with one of the following conditions: Lymphadenectomy can be considered as a staging procedure to better tailor adjuvant treatment (37,41).
EBRT is recommended to decrease vaginal recurrence since it has been shown to reduce the risk of local relapse (29,32,42,43). Observation is an option, especially for patients aged <60 years (44).
For p53abn FIGO 2009 stage IA without myometrial invasion cases, adjuvant treatment should be discussed on a case-by-case basis since specific data are missing. Lymphadenectomy for nodal staging purposes can be considered (45).

High-Intermediate-Risk Class
The incidence of recurrence after surgery alone is between 12 and 14% (29,32).
For those patients who underwent surgical nodal staging documenting negative nodes, VBRT is recommended to decrease vaginal recurrence, but no adjuvant therapy with close follow-up is an alternative acceptable option (4). In the case of substantial LVSI, EBRT can be considered in order to reduce the risk of pelvic and para-aortic nodal relapse (46). Similarly, cases displaying grade 3 tumors and/or substantial LVSI could benefit from adjuvant chemotherapy (31).
In patients for which lymph nodal status is unknown, VBRT is recommended for those patients who have LVSI negative, while EBRT is recommended for LVSI unequivocally positive to decrease pelvic recurrence (31,46). Systemic therapy is considered of uncertain benefit (31).

High-Risk Class
This risk class includes patients with the following characteristics: -FIGO 2009 stage I non-endometrioid (serous, clear cell, undifferentiated carcinoma, carcinosarcoma, mixed) with myometrial invasion, and with no residual disease -FIGO 2009 stage I p53abn endometrial carcinoma with myometrial invasion, with no residual disease.
There is no agreement on the role of lymphadenectomy in this risk class (4).
For this class, EBRT with concurrent and adjuvant chemotherapy or alternatively sequential chemotherapy and radiotherapy is recommended (29,31,(46)(47)(48). However, chemotherapy additional benefit is unclear for patients with clear-cell carcinomas. Chemotherapy alone can be an alternative option (49).

ADDITIONAL FEATURES
Additional prognostic risk factors such as L1CAM and mutations in exon 3 of CTNNB1 later emerged and demonstrated to better mark differences in terms of prognosis among the four classes (16,34,(50)(51)(52). Overall, three different prognostic profiles were delineated (see Table 2).
These additional features and, as a consequence, these profiles were not included in the most recent guidelines but were adopted in PORTEC-4a study to assign adjuvant treatment in the experimental arm (17).

CTNNB1
CTNNB1 gene encodes b-catenin protein, involved in regulation and coordination of cell adhesion and cell signaling.
In particular, within the copy number low group, CTNNB1 exon 3 mutation status was found prognostic for distant recurrence in EC (7).
CTNNB1 status helped distinguishing, within this class, a favorable group (CTNNB1-wild type) with a similar prognosis to POLEmut tumors, from an unfavorable group (CTNNB1mutant), with a similar prognosis to MMRd.

L1CAM
L1CAM is a 200 to 220 kDa membrane glycoprotein of the immunoglobulin superfamily and is crucially involved in processes of neurogenesis (56).
The established ≥10% threshold for positivity was based on the cutoff that best correlated with prognosis (57). It has been shown that patients bearing L1CAM-positive cancers have poorer disease-free and overall survival (51). L1CAM positivity was mainly, but not exclusively, found in intermediate-and high-risk cancers (13.2 vs 25.8% in low and intermediate, respectively) (51). Moreover, it was associated with histopathological high grade and increasing depth of myometrium infiltration (58).
Given the association with an overwhelming increase in the likelihood of distal or local recurrence and poor overall survival, its presence indicates the need for adjuvant treatment (51,59).

CONCLUSIONS
The traditional dualistic histopathologic classification that split EC into two groups, type I and type II cancer, is not more adherent to practical necessity of the clinicians. In recent years it has become increasingly clear that the traditional classification lacks reproducibility and yields heterogeneous molecular groups, hampering advances and implementation of precision medicine. This is particularly problematic for future clinical trials with targeted approaches that will demand inclusion of cancers with molecular similarities. The endometrial cancer classification proposed by TCGA would serve this purpose well, as it is based upon the combination of somatic mutational burden and somatic copy number alterations. Moreover, several publications on large and clinically well annotated (trial) cohorts have shown that surrogate IHC markers can be utilized for a TCGA-inspired molecular classification in routine surgical pathology, without the need for extensive sequencing. These surrogate markers have been extensively studied and show good performance. The prognostic value has been well established, with POLEmut EC having an excellent outcome and p53abn EC having the poorest clinical outcome, independent of risk group, type of adjuvant treatment, tumor type, or grade. This implies that de-escalation of adjuvant treatment for POLEmut EC patients should be explored, as is currently being done in the clinical PORTEC4a trial. Furthermore, recent data strongly suggest that the benefit for the addition of chemotherapy in the adjuvant treatment is limited to p53mut EC, which includes most serous cancers but also a significant portion of other histologic subtypes such as carcinosarcomas, thus suggesting an escalation of adjuvant treatment with chemotherapy combined with radiation when p53 mutation is detected. The implementation of molecular classification into clinical classification has the potential to serve in improving patient management by reducing over-and undertreatment (60). The use of this novel classification in routine clinical practice and future trial designs should be encouraged. Currently, one trial (PORTEC-4a) is ongoing to determine whether adjuvant treatment can be based on a molecular-integrated risk profile rather than standard clinicopathological risk factors in high-intermediate-risk EC patients. The preliminary report of the first 50 patients enrolled showed that molecular assessment is feasible, but patients' acceptance rate was not completely satisfactory (around 35%) (17). Nevertheless, possible technical limits such as the need of assay harmonization as well as the lengthening of reporting times should be addressed.