WHAT IS ALREADY KNOWN ON THE TOPIC

Pre-operative staging and especially diagnostic tumor size estimation defines the therapeutic algorithm in cervical cancer. Patients eligible for radical surgery with a therapeutic intent are mainly those with International Federation of Gynecology and Obstetrics (FIGO) 2018 stage IB2. Patients with tumors between 2 and 4 cm may be at a higher predisposition to undergo adjuvant treatment after surgery, thus increasing patient morbidity.

WHAT THIS STUDY ADDS

This study shows that there is excellent agreement on estimates of tumor size by MRI when compared with final pathology tumor size. In specifically evaluating patients who underwent a pre-operative MRI, the accuracy was high (83.2%) compared with final pathology. Concordance between physical examination and tumor measurement by MRI was also high (91.1%). We also noted that most patients with tumors >3 cm will undergo adjuvant therapy, potentially predisposing these patients to higher rates of associated complications.

HOW MIGHT THIS STUDY AFFECT RESEARCH, PRACTICE OR POLICY

These findings suggest that future work should determine strategies to explore which patients are ideal candidates for surgery alone in order to avoid potential adverse events from surgery followed by chemotherapy and radiation. In addition, use of MRI remains the ideal tool for assessing patients with early cervical cancer when determining the approach to therapy.

INTRODUCTION

Cervical cancer is the fourth most common cancer in women worldwide, with an estimated 604 000 new cases and 342 000 deaths per year according to the WHO.1 The American Cancer Society estimates about 13 820 new cases of invasive cervical cancer and 4360 deaths in the USA in 2024.2 Diagnosis and prognosis depend on the stage, which is described according to the 2018 International Federation of Gynecology and Obstetrics (FIGO) classification update. Under this most recent classification, stage IB is subdivided into stage IB1 (≤2 cm), IB2 (>2 and ≤4 cm), and IB3 (>4 cm).3 In patients with intermediate risk factors (combination of increased tumor size, lymphovascular space invasion (LVSI) or stromal invasion) adjuvant therapy after radical surgery is routinely recommended in the form of radiation, while if patients present with high risk factors (positive pelvic lymph nodes, parametrial involvement, or a positive surgical margin) adjuvant chemoradiation should be considered.4–6

In patients with early cervical cancer, using more than one treatment modality results in a substantial increase in the number and severity of adverse events, such as lymphedema, sexual dysfunction, urinary complications, diarrhea or constipation, and bowel obstruction.7 To this end, in patients considered ideal candidates for surgery, one should aim to avoid a combination of treatment, and thus an optimal pre-operative evaluation is essential to select the most suited therapy. Patients with tumor size >2 cm and ≤4 cm (FIGO 2018 stage IB2) are most at risk for combined treatment in the form of radical surgery and adjuvant chemotherapy and/or radiation depending on high-risk factors, especially in cases of lymph node involvement.6 8 Such patients have a higher propensity for poor estimation on pre-operative tumor size evaluation, and currently there are limited data on pre-operative evaluation and pathological size assessment.9

There is currently a gap in knowledge pertaining to the accuracy of pre-operative evaluation, including physical examination and imaging studies, and the relationship to tumor size on final pathology. In addition, there seems to be a paucity of data on rates of adjuvant treatment in a select group of patients (FIGO 2018 stage IB2) after radical hysterectomy, particularly when stratified based on tumor size within FIGO 2018 stage IB2. A specific interest aims at the question of whether patients with larger tumors should be referred to chemoradiotherapy directly, particularly if there is a higher rate of risk factors that dictate adjuvant treatment. The goal of this study was to determine the accuracy of pre-operative tumor size assessments by physical examination and imaging techniques to estimate post-operative tumor size on final pathology. Our secondary endpoints included distributions of risk factors and rates of adjuvant therapy based on pre-operative assessment with a subgroup analysis of patients with tumor size ranging from 3–4 cm.

METHODS

The patients included in this retrospective study were collected from three separate databases: the MD Anderson Cancer Center Radical Hysterectomy Database, the SUCCOR Study Group Database, and the Multi-institutional Database LATAM (encompassing Latin America and Europe). Patients with FIGO 2018 stage IB2 cervical cancer on pre-operative evaluation who underwent radical hysterectomy with a therapeutic intent were included. Patients included underwent surgery between April 17, 1991 and July 2, 2019 at MD Anderson, between January 2, 2013 and December 30, 2014 in the SUCCOR group, and between March 31, 2006 and November 27, 2017 in the LATAM cohort. For patients who underwent surgery prior to 2018, we reclassified these to conform to the FIGO 2018 cervical cancer staging. We included squamous carcinoma, adenocarcinoma, and adenosquamous carcinoma, any tumor grade, and pre-operative evaluation through clinical evaluation and/or imaging (ultrasound, MRI, CT, or positron emission tomography/CT (PET/CT)). Patients with incomplete documentation of treatment, pregnant patients, patients who received neoadjuvant chemotherapy and/or radiotherapy, patients who underwent prior conization or with missing documentation of post-operative tumor size were excluded.

For our primary objective, we used the methods described by Bland and Altman10 and estimated bias with a 95% confidence interval (95% CI). These methods were used to compare pre-operative tumor size by various imaging methods and physical exam with post-operative tumor size. Patients may have had pre-operative tumor size estimated to be ≤2 cm by either physical examination or one of the imaging methods, but any such patient was included if they also had pre-operative tumor size estimated to be >2 cm by another method. In addition, patients who had tumors >4 cm on imaging studies but with physical examination demonstrating tumors ≤4 cm were also included, as consideration for surgery was primarily dictated by physical examination. Tumor size was defined as the maximum diameter both at imaging and by pathology.

The Protection of Patient Confidentiality was preserved and no patient identifying information was used in the publication of our findings. All data were collected and managed using Research Electronic Data Capture (REDCap) tools hosted at MD Anderson.11 12 Study personnel at each institution were blinded to the data from other institutions. De-identified data files from each institution were transferred to the study statistician at MD Anderson using a secure file transfer application within REDCap.

RESULTS

A total of 714 patients were evaluated and 675 patients met the eligibility criteria (SUCCOR=350, LATAM=250, and MD Anderson=75). The median age was 46 years (range 22–82), and the median body mass index was 25.6 kg/m2 (range 15.1–68). The most common histologic subtype was squamous carcinoma (68%), and most patients had either grade 2 (53%) or grade 3 (31%) disease. The mean (SD) depth of invasion was 14.3 (8.56) mm and 50% (n=338) of patients did not have LVSI. The rate of positive lymph nodes in the entire population was 17% (n=113). A total of 58 (9%) patients had parametrial involvement and 3% (n=15) had vaginal involvement on final pathology. Surgical approach was by laparotomy in 60% (n=404) of patients and minimally invasive in 40% (n=271) of patients. Demographic data are shown in Table 1.

Table 1

Demographic and clinical characteristics

Overall imaging modalities used were MRI 52% (n=352), ultrasound 21% (n=140), CT 5% (n=32), and PET/CT 1% (n=10). Of note, some patients had more than one imaging modality. Overall, 413 patients had at least one type of imaging while a total 262 patients had tumor size estimated only by physical examination. The median pre-operative tumor size by the respective approaches was as follows: physical examination 3 cm (range 2–4); MRI 3 cm (range 1.0–4.8); ultrasound 3 cm (range 1.5–4.8); CT scan 3.1 cm (range 2–5); PET/CT 2.3 cm (range 1.4–4.1).

Bland-Altman Analysis

The bias is calculated as the average difference between the pre-operative and final pathology estimate of tumor size. Thus, if the bias is positive, the pre-operative evaluation overestimates the tumor size. The results of the Bland-Altman analysis showed that pre-operative physical examination, MRI, ultrasound, and CT all overestimated the tumor size. However, only the biases found for physical examination (p<0.0001) and MRI (p=0.0102) were statistically significant. Specifically for ultrasound, there were 140 patients (bias 0.085, 95% CI 0.04 to 0.22, p=0.20) (online supplemental table 1).

In specifically evaluating patients who underwent a pre-operative MRI, a total of 293 (83.2%) with tumor size 2–4 cm by MRI had the same range of tumor measurement on final pathology. No patient with tumor size by MRI between 2–4 cm had tumor >4 cm by final pathology. When classifying tumor size measurements into three categories (<2 cm, 2–4 cm, >4 cm), the 352 patients with tumors estimated by MRI agreed with estimates by pathology in 303 (86.1%) patients. Similarly, when evaluating agreement of physical examination with tumor size by MRI, we found that findings were consistent in 319 (91.1%) patients. These data are shown in Table 2. In addition, when evaluating concordance of estimated tumor size by physical examination with size based on final pathology, we found this to be 80.6% (all histologies) and 76.3% (adenocarcinoma and adenosquamous carcinoma).

Table 2

Tumor size on pathology and pre-operative MRI

A total of 343 (51%) patients received adjuvant treatment, with the majority of these (186 of 343, 54.2%) receiving chemoradiation, followed by radiation alone (152 of 343, 44.3%). The median tumor size on pathology among 338 patients who had adjuvant chemoradiation or radiation alone was 3.0 cm (range 0.1–6.5).

Subgroup Analysis

There were 340 (50.4%) patients who had tumor size on pathology ≥3 cm (including those with tumor size >4 cm), and 207 (60.9%) of these received adjuvant therapy. Of the 340 patients who had tumor size ≥3 cm, 111 (32.7%) had chemoradiation, 92 (27.1%) had radiation only, and four (1.2%) had chemotherapy only. There were 335 patients with tumor size on pathology <3 cm, of which 136 (40.6%) received adjuvant treatment. One (0.3%) had chemotherapy only, 60 (17.9%) had radiation only, 75 (22.4%) had chemoradiation, and 199 (59.4%) had no adjuvant treatment. There were 199 patients who had tumor size on pre-op MRI ≥3 cm, and 127 (63.8%) had adjuvant therapy. Sixty-one (30.7%) had radiation only, and 66 (33.2%) had chemoradiation. There were 202 (29.9%) patients who had tumor size on pathology ≥3.5 cm, and 137 of these patients (67.8%) had adjuvant therapy. Sixty (29.7%) had radiation only, and 75 (37.1%) had chemoradiation. Two (1%) had chemotherapy only.

There were 316 (46.8%) patients with tumor size 3–4 cm on pathology. The different distributions of LVSI, lymph node status, and parametrial involvement in tumors 2–2.99 cm and 3–4 cm are shown in Table 3. In tumors of 3–4 cm, LVSI was present in 144 (49%) of 294 cases, lymph nodes were positive in 66 (21%) of 316 cases, and parametrial invasion was detected in 35 (11%) of 315 cases. The rate of LVSI in tumors 2–2.9 cm and 3–4 cm was 49% for both groups. Similarly, rates of parametrial involvement were 7% and 11%, respectively (p=0.098). There was a significantly higher incidence of positive lymph nodes with increasing tumor size (2–2.99 cm, 13% (29/222) vs 3–4 cm, 21% (66/316), p=0.022).

Table 3

Risk factor distribution between subgroups (tumor size on pathology)

DISCUSSIONSummary of Main Results

Overall, our study focused on accuracy of pre-operative size estimation, especially performed by physical examination and MRI, which are the two most often adopted assessments tools worldwide. In our cohort, there was a high (83.2%) concordance between MRI and final pathology in patients with tumor size 2–4 cm. In addition, agreement of physical examination with tumor size by MRI was also high (91.1%). As it pertains to rates of adjuvant treatment, our study found that 61% of the patients with tumor size ≥3 cm on final pathology received adjuvant therapy after undergoing radical hysterectomy with a curative intent. In addition, in patients who had a tumor size on final pathology ≥3.5 cm, more than two thirds (67.8%) underwent adjuvant therapy. Lastly, with tumor size increase there was a significant increase of risk factors, such as rates of positive lymph nodes even within the same FIGO stage (IB2).

Results in the Context of Published Literature

Pathologic tumor size is an independent prognostic factor for cause-specific survival. Specifically, when compared with stage IB1 disease, stage IB2 disease is associated with a nearly two-fold increased risk of cervical cancer mortality (HR 1.98, p<0.001).13 Consequently, international guidelines are increasingly emphasizing the role of cross sectional imaging in the staging of cervical cancer.14 Physical examination has been reported to have poor accuracy when determining tumor size in cervical cancer. In a study by Qin et al, the investigators retrospectively analyzed 818 patients with operable stage IB-IIB cervical cancer determined by pelvic examination. The analysis showed that concordance diminishes as stage increases: 85.4%, 77.4%, 35.3%, and 20.5% for FIGO 2009 stage IB1, IB2, IIA, and IIB, respectively.15 In an effort to evaluate tools for increased accuracy of pre-operative imaging, fusion techniques are being explored and future studies will likely shed light into how these novel approaches may impact patient selection.16

It has long been recognized that using more than one treatment modality results in a substantial increase in treatment complications, such as lymphedema, sexual dysfunction, urinary frequency, diarrhea or constipation, and bowel obstruction.7 In the prospective STAR trial, patients with stage IB to IIA cervical cancer with pathological risk factors were randomized to adjuvant radiotherapy, concurrent chemoradiation or sequential chemoradiation after radical hysterectomy, with the primary aim to evaluate disease-free survival among three radiotherapy approaches.17 A total of 291 of 298 (97.7%) patients in the chemoradiation group and 271 of 303 (89.4%) in the radiation alone group had at least one adverse event of any grade (p<0.001). When considering grade 3 or 4 adverse events, the radiation alone group had a 12.9% complication rate compared with 28.5% in the chemoradiation group. More recently, intensity-modulated radiotherapy has been used after radical hysterectomy. The RTOG1203 trial evaluated patient-reported outcomes using intensity-modulated radiotherapy after surgery in patients with cervical and endometrial cancer. The investigators found that between-arm comparison of patient-reported high-grade adverse events, patients receiving intensity-modulated radiotherapy experienced fewer gastrointestinal adverse events (18.2% difference, p=0.01).18

Trying to reduce combined treatment, Hwang et al retrospectively evaluated patients with FIGO 2018 stage IB2 cervical cancer measured by MRI.19 The proportion of patients who had positive LVSI, deep stromal invasion, lymph node metastasis or involvement of parametrium were significantly higher in patients with an estimated tumor size ≥2.7 cm. The authors concluded that beyond a size of 2.7 cm on pre-operative MRI, chemoradiation rather than surgery should be recommended, thus avoiding the combined approach. In this context, our study also provides further evidence that a significant (67.8%) number of patients will undergo adjuvant treatment after radical hysterectomy for larger tumors (≥3.5 cm), thus providing further evidence that future studies are needed to more accurately select ideal candidates for surgery in patients with FIGO 2018 stage IB2.

Strengths and Weaknesses

One of the strengths of our study is its multicenter approach with a large number of patients from three cohorts from specialized oncological centers. Furthermore, the data sources were centers from Europe, South America, and North America providing a global real-world insight on diagnostic tools used in the evaluation of patients prior to radical hysterectomy.

However, we recognize there are a few limitations in our study. The retrospective nature of our data collection may be prone to inconsistencies among countries as to how data were interpreted and entered. In addition, the study spans a broad time period when there might have been variations in terms of imaging quality, indications for adjuvant treatment, and recommendations rendered. Similarly, the databases used encompassed a specific time frame that reflects a gap in more recent data. However, given the retrospective nature of the study, we provided an analysis on the data that were available at the time of study design. The indications for radical hysterectomy or indications for adjuvant treatment were not outlined in the source documents. Additionally, the lack of standardized pathological evaluation also imposes a potential bias when considering the granular evaluation of tumor sizes. We also could not provide details as to the exact location of the tumor within the cervix as this may have an impact on determination of tumor size by physical examination. Variations in radiological interpretation of the different imaging modalities hinder potential reproducibility of the findings. Similarly, given the retrospective nature of the study, we were not able to capture details of the MRI protocols used, specifically if diffusion-weighted imaging was utilized. Furthermore, the overestimation of tumor size by MRI of 0.1 cm and by physical examination of 0.4 cm by the Bland-Altman analysis, even if statistically significant, does not have a clinical impact.

Implications for Practice and Future Research

The decision-making process for treatment of early-stage cervical cancer requires accurate pre-operative tumor size estimation. Our study highlights the need for continued evaluation of novel technologies in pre-operative imaging and appropriate patient selection to avoid multi-modality therapy. Furthermore, we also show that patients with cervical tumors ≥3 cm are at higher risk for adjuvant treatment after radical surgery. Future studies should aim to prospectively evaluate accuracy of imaging modalities such as pelvic MRI, ultrasound, and PET/CT imaging to select patients who are truly optimal for radical surgery.

CONCLUSION

Our study showed that both physical examination and routine pre-operative imaging with MRI provide adequate estimates of tumor size in patients with early-stage cervical cancer. In addition, in patients with tumors measuring 3–4 cm, the majority will undergo adjuvant chemotherapy and radiation. Further studies are needed to prospectively evaluate patients who are considered ideal candidates for surgery alone to avoid the combined treatment modalities.

Data availability statement

In accordance with the journal’s guidelines, we will provide our data for independent analysis by a team selected by the Editorial Team for the purposes of additional data analysis or for the reproducibility of this study in other centers if such is requested. Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by the Institutional Review Board (IRB) of the MD Anderson Cancer Center, IRB number 2023-0374, OHRP IRB Registration Number: IRB00000121. Participants gave informed consent to participate in the study before taking part.