- Email: [email protected]
Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer
YGYNO-976622; No. of pages: 5; 4C: Gynecologic Oncology xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer Jason R. Lefringhouse a,⁎, Jeffrey W. Elder a,1, Lauren A. Baldwin a, Rachel W. Miller a, Chris P. DeSimone a, John R. van Nagell Jr a, Luis M. Samoyoa b, Dava S. West b, Emily V. Dressler c, Meng Liu c, Frederick R. Ueland a a b c
Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Kentucky, Markey Cancer Center, Lexington, KY, United States Department of Pathology, University of Kentucky, Markey Cancer Center, Lexington, KY, United States Division of Cancer Biostatistics, University of Kentucky, Markey Cancer Center, Lexington, KY, United States
a r t i c l e
i n f o
Article history: Received 15 November 2016 Received in revised form 30 January 2017 Accepted 31 January 2017 Available online xxxx Keywords: Endometrial cancer Intraoperative Surgical staging Lymph node Mayo Clinic algorithm, prospective
a b s t r a c t Objectives. Prospectively validate an intraoperative surgical staging algorithm to stratify patients with early endometrial cancer by risk of lymph node metastasis. Methods. Subjects with endometrial cancer clinically confined to the uterus were prospectively enrolled at an academic cancer center between Jan 2012 and Jun 2015. Study participants were stratified intraoperatively into two groups based on risk of nodal involvement using cell type, tumor grade, myometrial invasion, and tumor size in accordance with an established protocol from the Mayo Clinic. Low risk (LR) subjects received extrafascial hysterectomy with bilateral salpingo-oophorectomy; high risk (HR) patients received complete surgical staging including bilateral pelvic and para-aortic lymphadenectomy. Results. Of the 200 subjects enrolled, 194 were eligible for analysis. The algorithm identified 132 (68%) HR and 62 (32%) LR cancers. Of the HR subjects, 126 had lymphadenectomy performed with 14 (11%) positive for nodal metastases. Five HR subjects experienced disease recurrence. Of the 62 LR cancers, two patients developed disease recurrence. Ten LR cancers were upgraded to HR on final pathology due to lesion size (6) and grade (4). None of these patients experienced disease recurrence. The algorithm demonstrated 90% sensitivity (18/20) and 36% specificity (62/174) as determined by positive lymph nodes and/or disease recurrence. Conclusions. Intraoperative assessment of early endometrial cancer can be used to determine the extent of surgical staging. The studied algorithm has low specificity and modifications are necessary to better match the surgical procedure to the risk of metastatic cancer. Published by Elsevier Inc.
1. Introduction Endometrial cancer (EC) is the most common gynecologic malignancy in the United States, and the incidence continues to rise. In 2016, there are expected be 60,050 new cases of EC with an estimated 10,470 deaths [1]. The majority of women with EC present with early stage disease where survival rates exceed 80% following primary surgery [2]. In the United States, standard surgical staging for women with EC includes extrafascial hysterectomy, bilateral salpingo-oophorectomy, and lymphadenectomy. In 2014, the Society of Gynecologic Oncology Clinical Practice Endometrial Cancer Working Group
⁎ Corresponding author at: Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Kentucky, Markey Cancer Center, 800 Rose Street, Lexington, KY 40536, United States. E-mail address: [email protected] (J.R. Lefringhouse). 1 Current affiliation: Division of Gynecologic Oncology, Greenville Hospital System Cancer Institute, Greenville, SC.
reported: 1) patients with grade 1–2 endometrioid tumors, b 50% myometrial invasion, and tumor of 2 cm or less are at low risk for recurrence and may not require a pelvic and para-aortic lymphadenectomy, and 2) lymphadenectomy may alter or eliminate the need for adjuvant therapy and its associated morbidity [3]. Today, there remains disagreement among specialists both on the necessity and extent of surgical staging for women with early EC [2,4–6]. The Mayo Clinic has published extensively on the surgical management of early EC, defining a subgroup of women at very low risk for lymph nodal involvement in which lymphadenectomy can be omitted. The authors developed an institutional algorithm that directs surgical treatment based on pathologic intraoperative consultation (IOC). Intraoperative stratification is based on the risk of nodal involvement: low risk (confined to corpus, grade 1 or 2 endometrioid type, myometrial invasion ≤50%, tumor diameter ≤ 2 cm), or high risk (does not meet low risk criteria). The low risk (LR) group is not required to undergo lymph node sampling or lymphadenectomy, while the high risk (HR) group receives bilateral pelvic and para-aortic lymphadenectomy [5,7]. In 2008,
http://dx.doi.org/10.1016/j.ygyno.2017.01.032 0090-8258/Published by Elsevier Inc.
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
2
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
the authors published results on 422 women treated with this algorithm. Of the 112 in the LR group not requiring lymphadenectomy, 22 (20%) had lymphadenectomy and all nodes were negative for metastases. In the HR group, 22% had metastatic nodal involvement identified by lymphadenectomy. The investigators determined that the five-year recurrence-free survival was 98% in low risk women and 54% in high risk patients (p b 0.001) [8]. Despite these encouraging reports, further study is needed to determine how the intraoperative algorithm performs outside the unique Mayo Clinic setting where the intraoperative pathology processing differs substantially from other institutions. This prospective investigation of women with early endometrial cancer was undertaken to evaluate the performance of the Mayo Clinic algorithm and its concordance with final pathology when using targeted intraoperative frozen section analysis. 2. Materials and methods Study participants who met eligibility criteria were consecutively enrolled and underwent surgical treatment at the University of Kentucky between January 2012 and June 2015. Institutional review board approval was obtained and the trial was registered on ClinicalTrials.gov (NCT01512810). Eligible participants had histologically confirmed EC clinically confined to the uterus. The use of preoperative imaging was not required but could be used at the discretion of the treating physician. Eligible subjects had a GOG performance status of 0, 1, or 2, and were surgical candidates for a complete hysterectomy, bilateral salpingo-oophorectomy, and pelvic and para-aortic lymphadenectomy. Subjects were included if a previous diagnosis of invasive cancer or chemotherapy was N5 years prior to enrollment; those with prior vaginal, pelvic or abdominal irradiation were excluded. The treating surgeon and patient collectively determined whether the operation was via laparotomy or a robotic-assisted laparoscopic approach. Laparoscopy alone was not offered because the participating surgeons prefer to perform lymphadenectomy with robotic-assistance. When surgically removed, the uterus and cervix were promptly sent to surgical pathology for intraoperative evaluation according to the IOC algorithm. The uterus was bivalved in the coronal axis. Visible endometrial lesions were measured in three dimensions with the longest dimension defined as the primary tumor diameter. At least one representative frozen section of the tumor was performed and a systematic intraoperative risk assessment was completed. If multiple lesions were noted, the largest lesion was measured and recorded. Histologic type was classified according to World Health Organization criteria. Architectural grading was based on degree of glandular differentiation according the FIGO guidelines and assignment of stage utilized the 2009 FIGO staging system [9]. In accordance with the Mayo Clinic criteria, the LR group was defined by the presence of all of the following criteria: endometrioid cell type with disease confined to the uterine corpus, any grade without myometrial invasion, grade 1 or 2 tumors with myometrial invasion less than or equal to 50%, and primary tumor diameter ≤ 2 cm. Low risk subjects underwent hysterectomy with bilateral salpingo-oophorectomy, but were not required to have pelvic and para-aortic lymphadenectomy. The remaining subjects were classified as HR and were surgically staged with bilateral pelvic (right and left external iliac, internal iliac, obturator, common iliac) and bilateral para-aortic lymphadenectomy to the level of the inferior mesenteric artery. The study employed a prospective non-randomized stratified observational design to estimate 2-year recurrence-free survival (RFS) rates. Sample size was determined a priori, using retrospective data from the University of Kentucky Endometrial Cancer Database; we predicted 2year RFS rates to be 92% in the low-risk and 82% in the high-risk subgroup. With the assumption that 60% of subjects would be categorized as low risk during IOC, 167 eligible patients would provide 80% power to estimate the 2-year RFS rate within ±5.8% for low risk and ±9.8% for high risk based on a two-sided exact 95% binomial confidence interval for a single proportion. If low-risk subgroup rates of 55% or 50% were observed,
the total sample size would be increased to 182 or 200 respectively to maintain similar precision. Thus, the final sample size was 200. Once subjects were classified as LR or HR, statistical analysis included Fisher's exact test and 2-sample t-tests to assess for differences between risk statuses for categorical and continuous variables respectively. Discordance between IOC and final path reports were calculated with corresponding Kappa statistics. Differences were considered statistically significant at p b 0.05. Statistical analyses were conducted using the SAS version 9.4 software package (SAS Institute, Inc.: Cary, NC). 3. Results A total of 200 subjects with EC were consecutively enrolled and 194 were eligible for analysis. Of the six ineligible subjects, three did not meet inclusion criteria, two declined surgery, and one had surgery at another facility (Fig. 1). The mean lymph node count for the 126 high risk patients who had lymph nodes removed was 15.4 (range 0–44). The mean lymph node count for the patients who had a completed pelvic and para-aortic lymphadenectomy was 17.7 (range 6–44). 3.1. Findings related to risk groups Based on intraoperative evaluation, 32% (62/194) were LR and 68% (132/194) were HR. The clinical and pathological characteristics are shown in Table 1. Five subjects in the LR group not requiring lymphadenectomy by protocol received at least partial lymphadenectomy. All lymph nodes were negative on final pathology, but one patient developed disease recurrence at 18 months after surgery. In all, two subjects (3%) in the LR group developed recurrent disease. They were the only patients in the LR group to receive additional therapy following surgery. Fourteen of 132 (10%) subjects in the HR group had positive lymph nodal metastases on final pathology. Intraoperatively, all were grade 2 or 3 tumors or had a lesion size N3 cm. Three patients had a lesion size of 2–3 cm. There were no lymph node metastases identified for grade 1 tumors with a lesion size ≤3 cm. Five of 14 (36%) had cervical involvement (Table 2). Of the 132 subjects in the HR group, 32% received either adjuvant radiation (11%), chemotherapy (14%), or both (7%). All HR subjects with lymph node metastases received adjuvant therapy. Of the seven subjects who developed disease recurrence, five were HR including one with positive lymph nodes and two with cervical involvement on final pathology. Two subjects with recurrence were in the LR cohort (Table 3). One LR subject was initially classified as HR by tumor size because the lesion grossly appeared to replace the entire surface of the endometrium. Lymph node dissection was initiated (seven negative para-aortic nodes), but a representative frozen section identified only complex atypical hyperplasia and the lymph node dissection was discontinued. The final pathology revealed a grade 1, endometrioid adenocarcinoma with a single focus of invasive disease confined to the endometrium. She developed a pelvic recurrence at two years and was treated with pelvic radiation; a supra-clavicular lymph node recurrence 10 months later was treated with chemotherapy; the patient is alive 29 months after completion of therapy. The second subject with a LR recurrence had a 1.5 cm, grade 1 endometrioid adenocarcinoma with 10% myometrial invasion on final pathology. A vaginal cuff recurrence at seven months was treated with whole pelvic radiation and vaginal cuff brachytherapy. The patient is alive and without definitive progression 26 months after her recurrence. During the mean follow up of 15.3 months (range 1–43), six subjects died: four were disease related and two were from causes unrelated to their cancer diagnosis. At this time, there is no difference in recurrencefree survival between the HR and LR groups on Kaplan-Meier analysis. 3.2. Findings related to individual algorithm components The discordance by risk group between IOC and final pathology was 5.7% (11/194, K = 0.861, p b 0.0001). Sixteen percent (10/62) of subjects
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
3
Fig. 1. CONSORT diagram summarizing study flow.
initially classified as LR were HR on final pathologic evaluation. Six were due to changes in size of lesion, all of which were grade 1 tumors and 4 were reclassified from endometrioid cell type intraoperatively to serous post-operatively. Disparities in myometrial invasion and grade were not responsible for any changes in risk group. None of the ten patients reclassified from LR to HR elected to undergo subsequent lymphadenectomy and none have recurred. The remaining patient (1/132) was intraoperatively considered HR but later deemed LR on final pathology based on re-measurement of tumor size. When considering each factor on univariate analysis, there was a high degree of concordance between IOC and final pathology for risk stratification (kappa = 0.861, p b 0.0001, as well as the individual risk components of tumor size categories (kappa = 0.8629), grade (kappa = 0.7078), histology (kappa = 0.6561) and myometrial invasion (kappa = 0.5628) (p b 0.0001 for each). Regarding myometrial invasion, 3.7% (6/164) of subjects with ‘none’ or ≤50% invasion were changed to N50%, while 17% (5/30) where changed from N50% to ≤50% invasion. The overall discrepancy rate for myometrial invasion when divided into these two categories (≤50% and N50%) was 5.7% (11/194). These discrepancies did not result in risk group reclassification. The discordance rate for IOC lesion size to final pathology was 4.3% (8/184). Of the 62 subjects with IOC tumor size ≤ 2 cm, six were upgraded to N2 cm, while two of 125 subjects with size N2 cm were downgraded. A similar discrepancy rate of 4.9% (9/184) would have
been observed if the cutoff for lesion size were changed to ‘any’ for grade 1 and b 3 cm for grade 2 tumors. For tumor grade, 5.7% (9/158) of grade 1–2 intraoperative tumors were grade 3 on final pathology, while 5.7% (2/35) grade 3 tumors were downgraded to grade 1–2 for a discrepancy rate of 5.7% (11/ 193). These grade changes did not result in a change of risk group. 4. Discussion There is ongoing debate over the utility and extent of lymphadenectomy for the surgical treatment of early stage EC. A number of key factors influence the discussion, including: the specifics of preoperative and intraoperative risk stratification, operative morbidity and mortality, cost of treatment, and patient survival. Even if patient survival is not affected by lymphadenectomy in women with early EC, tailoring the operation to match the extent of disease is an appealing concept to reduce the morbidity of treatment. The IOC algorithm is intended to segregate early endometrial cancers into LR and HR groups during surgery to give surgeons an intraoperative surrogate for factors associated with the risk of extra-uterine involvement on final pathology [10]. Twenty-five years ago, Schink and coauthors proposed tumor size as an important component of EC risk stratification [11]. More recently, depth of myometrial invasion, tumor grade and cell type have been added as significant factors [10–12]. Mariani and colleagues at the
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
4
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
Table 1 Patient demographics and disease related parameters by intraoperative risk for lymph node involvement. Intraoperative group
Number of patients Age at diagnosis (years) BMI Underweight + normal (b25) Overweight (25–30) Obese I (30–35) Obese II (35–40) Obese III (40+) Race White Black or African American Pre-op FIGO Grade, N (%) Grade 1 Grade 2 Grade 3 Grade unknown, not stated Pre-op histologic subtype Adenocarcinoma, NOS Endometrioid Serous Clear cell Adenosquamous Carcinosarcoma Undifferentiated Not applicable Lymphadenectomy Yes No Type of hysterectomy Total abdominal Robot-assisted laparoscopic
High risk
Low risk
p-Value
132 61.4 (10.0)
62 56.3 (9.4)
0.0009
17 (13.0%) 23 (17.6%) 34 (26.0%) 24 (18.3%) 33 (25.2%)
11 (18.0%) 7 (11.5%) 12 (19.7%) 13 (21.3%) 18 (29.5%)
0.5772
128 (97.0%) 4 (3.0%)
62 (100.0%) 0 (0.0%)
0.3082
66 (50.0%) 33 (25.0%) 30 (22.7%) 3 (2.3%)
47 (75.81%) 13 (21.0%) 2 (3.2%) 0 (0.0%)
9 (6.8%) 109 (82.6%) 8 (6.1%) 2 (1.5%) 1(0.8%) 1(0.8%) 1(0.8%) 1(0.8%)
4 (6.4%) 58(93.6%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%)
126 (95.5%) 6 (4.6%)
5 (8.1%) 57 (91.9%)
96 (62.5%) 36 (37.5%)
44 (59.0%) 18 (40.9%)
0.0003
0.4394
b0.0001
0.7987
p-values were calculated from t-test for continuous variables and chi-squared test and Fisher's exact test for categorical variables; NOS - not otherwise specified.
Mayo Clinic combined these features into a specific intraoperative algorithm to be used in the surgical triage of early endometrial cancers. They published a 5% rate of lymph node involvement and a 96% recurrencefree survival in the LR group [7]. Other investigators have studied the Mayo Clinic algorithm. Convery and coauthors completed a two center, retrospective chart review of 110 subjects with early EC and reported that 2.4% of women with LR cancers developed disease recurrence [13]. Vargas and co-investigators utilized Surveillance, Epidemiology, and End Results (SEER) data from 19,329 women to confirm that the LR group by Mayo criteria had a 1.4% risk of lymph node metastasis
compared to 6.4% with HR features [14]. In a community setting, Bell and coauthors retrospectively evaluated the outcomes of 179 patients found to be LR by intraoperative assessment and surgically managed without lymphadenectomy and reported a 1.7% recurrence risk and five year overall survival of 95.8% [15]. The findings of our prospective investigation confirm that the Mayo Clinic algorithm can be successfully employed at other institutions. In agreement with previous reports, we find a 3% (2/62) recurrence rate for the LR group, one distant (pelvic sidewall and supra-clavicular node) and one local (vaginal cuff). The most challenging aspect of implementing the Mayo Clinic algorithm is standardizing the processing of surgical pathology specimens. In brief, the Mayo clinic has a robust, intraoperative processing capability where multiple frozen sections are performed (often the entire specimen) and the final pathologic diagnosis is rendered immediately. In rare instances, the final diagnosis is delayed for further processing with immunohistochemical staining. This practice is not uniform in the United States. At the other extreme, facilities without frozen section analysis have attempted to circumvent this obstacle by combining the preoperative biopsy with intraoperative tumor dimensions [13]. Evaluating our specimen processing results, the risk category differed from IOC to final pathology in 5.7% of patients analyzed. Ten of the 62 (16%) LR subjects were upgraded, six for lesion size and four for grade (all endometrioid to grade 3 serous cell type). Three with serous histology on final pathology received adjuvant chemotherapy with carboplatin and paclitaxel. None of the ten received lymphadenectomy and none have recurred at a mean follow up of 15.3 months. Only one subject was downgraded on final pathology for tumor size. Thus, the IOC with targeted frozen section analysis compared favorably with final pathology in terms of individual risk factors and relevant clinical outcomes. To further minimize discordance, the authors recommend that the IOC include at least one frozen section (preferably two or more) with specific attention to intraoperative grade, cell type, depth of myometrial invasion, and cervical involvement. The IOC frozen section should confirm that the measured lesion is malignant and reassess tumor grade. Tumor grade should not be lowered even if it differs from the preoperative result, and a cancer diagnosis should not be overturned from a preoperative biopsy. Whenever possible, preoperative biopsies should be reviewed at the treating institution prior to surgery. The intraoperative algorithm demonstrated 90% sensitivity (18/20) and 36% specificity (62/174) for positive lymph nodes and disease recurrence. The algorithm performed well at identifying EC at HR of extra-uterine disease; however, by recommending lymphadenectomy for over two-thirds of the study population, the IOC criteria mis-
Table 2 Pathologic findings and outcomes for subjects with positive lymph nodes. Patient #
1 2 3 4 5 6 7 8 9 10 11 12 13a 14
Intraoperative pathology
Final pathology
Histology
Tumor size (cm)
Tumor grade
MI
Histology
Tumor size (cm)
Tumor grade
MI
Cervix involved
Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Serous
N3 N3 (2–3) N3 N3 N3 N3 N3 N3 (2–3) N3 (2–3) N3 N3
1 1 2 1 1 3 1 1 3 2 3 3 2 3
b50% N50% b50% b50% N50% b50% b50% N50% b50% b50% b50% b50% b50% b50%
Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Serous Endometrioid
N3 N3 (2–3) N3 N3 N3 N3 N3 N3 (2–3) N3 (2–3) N3 N3
1 2 2 3 2 3 2 2 3 1 2 3 3 3
b50% N50% b50% N50% b50% N50% b50% N50% b50% b50% b50% b50% b50% b50%
− + + − − + − + − + − − − −
All patients with positive lymph nodes were intraoperatively high risk per Intraoperative algorithm and post-operative pathology. MI - myometrial invasion. a One patient had both +LN and + disease recurrence.
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
5
Table 3 Pathologic findings and outcomes for subjects with disease recurrence. Patient #
Intraoperative evaluation Risk group
Histology
Tumor size (cm)
Tumor grade
Lymph-adenectomy Risk group
Histology
Tumor size (cm)
Tumor grade
Lymph node positive
Cervix involved
1 2 3 4 5 6 7a
High High High Low Low High High
Endometrioid Undifferentiated Adenosquamous Endometrioid Endometrioid Endometrioid Endometrioid
N3 N3 N3 ≤2 ≤2 (2–3) N3
2 3 3 1 1 3 2
Yes Yes Yes Yes No Yes Yes
Endometrioid Endometrioid Non-Endometrioid Endometrioid Endometrioid Endometrioid Serous
N3 N3 N3 ≤2 ≤2 (2–3) N3
3 3 3 1 1 3 3
− − − − N/A − +
+ − − − − + −
a
Final evaluation
High High High Low Low High High
One patient had both +LN and + disease recurrence.
categorized many LR cases as HR. In a recent SEER review, Vargas published similar results of the Mayo criteria with a test sensitivity of 85% (337/395) and specificity of 28% (4039/13,969). Similar to Mayo Clinic publications, Vargas reported that only 21% of subjects were classified as LR. To increase algorithm specificity, they proposed that the criteria be modified to selectively reduce the number of subjects included in the HR group while limiting an increase the number of false negative results [14]. The current study confirms that the Mayo criteria can be successfully implemented at other centers, and that published population-based reports are consistent with the findings of this prospective investigation. Our study demonstrated lymph node involvement in the LR group of 1.8%, 0%, and 0% compared to SEER data of 0.8%, 2.1% and 0% for grade 1, 2 and 3 tumors, respectively. Given that most cancers are categorized as high risk by the Mayo system, Vargas et al. suggested that the LR group be expanded to include grade 1 endometrioid tumors with b50% myometrial invasion regardless of lesion size, grade 2 tumors with lesion size b 3 cm, and grade 3 tumors with no myometrial invasion. In the Vargas SEER review, the LR group using this modified algorithm would have had a positive lymph node rate of 2.3%, 2.0% and 0% for grade 1, 2 and 3 tumors, respectively. This new algorithm would have a calculated test sensitivity of 63% and specificity of 66%. Extrapolating with our prospective data, the Vargas-proposed algorithm for the LR group would theoretically result in a positive lymph node rate of 2.2%, 0% and 0%, respectively. This change would increase specificity to 65%, decrease sensitivity to 75%, and decrease the number of women receiving pelvic and para-aortic lymphadenectomy from 68% to 38%. However, this change would potentially increase the number of false negative results from two to five. The strengths of this investigation are its prospective design, centralized pathology review, the consistent implementation of the IOC algorithm at a single institution, and both treatment and follow-up were performed by the same physicians. Our strategy of representative frozen section analysis during IOC is likely to be available at all medical centers with practicing gynecologic oncologists. A limitation of the study is the single institution design which limits generalizability. A second weakness is the short follow-up time of 15.3 months which excludes a meaningful survival analysis. The findings of this investigation indicate that the process of selective surgical staging with targeted frozen section for women with early endometrial cancer is reproducible and effective at identifying HR cancers with 90% sensitivity. There are two primary benefits to stratifying patients based on their risk of nodal metastases during surgery: 1) omission of lymphadenectomy in the LR group, and 2) tailoring adjuvant treatment for the HR group, possibly avoiding radiation when lymph nodes are histologically negative [15]. Future multi-institutional, prospective trials will determine if this strategy can improve the cost and safety of surgical treatment, without diminishing survival
outcomes. Since two-thirds of early EC are classified as high risk by the current IOC algorithm, we see an opportunity to thoughtfully expand the LR group to achieve a better clinical balance of surgical risk and treatment outcomes. Conflict of interest statement Emily Dressler PhD received grant funding through the University of Kentucky Markey Cancer Center Biostatistics Shared Resource Facility during the conduct of this study. NIH (NCI) P30 CA177558
References [1] R.L. Siegel, K.D. Miller, A. Jemal, Cancer statistics, 2016, CA Cancer J. Clin. 66 (1) (2016) 7–30. [2] H.C. Kitchener, E.L. Trimble, Intergroup ftECWGotGC. Endometrial cancer state of the science meeting, Int. J. Gynecol. Cancer 19 (1) (2009) 134–140. [3] William M. Burke JO, Mario Leitao, Emery Salom, Paola Gehrig, Alexander B. Olawaiye, Molly Brewer, Dave Boruta, Jeanine Villella, Tom Herzog, Fadi Abu Shahin, Endometrial cancer: a review and current management strategies: part I, Gynecol. Oncol. 134 (2) (2014) 8. [4] H. Kitchener, A. Swart, Q. Qian, C. Amos, M. Parmar, Efficacy of systematic pelvic lymphadenectomy in endometrial cancer (MRC ASTEC trial): a randomised study, Lancet 373 (9658) (2009) 125–136. [5] S.C. Dowdy, B.J. Borah, J.N. Bakkum-Gamez, A.L. Weaver, M.E. McGree, L.R. Haas, et al., Prospective assessment of survival, morbidity, and cost associated with lymphadenectomy in low-risk endometrial cancer, Gynecol. Oncol. 127 (1) (2012) 5–10. [6] P.B. Panici, S. Basile, F. Maneschi, A.A. Lissoni, M. Signorelli, G. Scambia, et al., Systematic pelvic lymphadenectomy vs no lymphadenectomy in early-stage endometrial carcinoma: randomized clinical trial, J. Natl. Cancer Inst. 100 (23) (2008) 1707–1716. [7] A. Mariani, M.J. Webb, G.L. Keeney, M.G. Haddock, G. Calori, K.C. Podratz, Low-risk corpus cancer: is lymphadenectomy or radiotherapy necessary? Am. J. Obstet. Gynecol. 182 (6) (2000) 1506–1519. [8] A. Mariani, S.C. Dowdy, W.A. Cliby, B.S. Gostout, M.B. Jones, Wilson TO, et al., Prospective assessment of lymphatic dissemination in endometrial cancer: a paradigm shift in surgical staging, Gynecol. Oncol. 109 (1) (2008) 11–18. [9] FIGO, Staging for carcinoma of the vulva, cervix, and corpus uteri, Int. J. Gynaecol. Obstet. 125 (2) (2014) 97–98. [10] H.M. Keys, J.A. Roberts, V.L. Brunetto, R.J. Zaino, N.M. Spirtos, J.D. Bloss, et al., A phase III trial of surgery with or without adjunctive external pelvic radiation therapy in intermediate risk endometrial adenocarcinoma: a gynecologic oncology group study, Gynecol. Oncol. 92 (3) (2004) 744–751. [11] J.C. Schink, A.W. Rademaker, D. Miller, J.R. Lurain, Tumor size in endometrial cancer, Cancer 67 (11) (1991) 2791–2794. [12] W.T. Creasman, C.P. Morrow, B. Bundy, H.D. Homesley, J. Graham, P. Heller, Surgical pathologic spread patterns of endometrial cancer. A gynecologic oncology group study, Cancer 60 (8 Suppl) (1987) 2035–2041. [13] M.M. AlHilli, A. Mariani, Preoperative selection of endometrial cancer patients at low risk for lymph node metastases: useful criteria for enrollment in clinical trials, J. Gynecol. Oncol. 25 (4) (2014) 267–269. [14] R. Vargas, J.A. Rauh-Hain, J. Clemmer, R.M. Clark, A. Goodman, W.B. Growdon, et al., Tumor size, depth of invasion, and histologic grade as prognostic factors of lymph node involvement in endometrial cancer: a SEER analysis, Gynecol. Oncol. 133 (2) (2014) 216–220. [15] C. Sharma, I. Deutsch, S.N. Lewin, W.M. Burke, Y. Qiao, X. Sun, et al., Lymphadenectomy influences the utilization of adjuvant radiation treatment for endometrial cancer, Am. J. Obstet. Gynecol. 205 (6) (2011) 562 (e1-.e9).
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
Contents lists available at ScienceDirect
Gynecologic Oncology journal homepage: www.elsevier.com/locate/ygyno
Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer Jason R. Lefringhouse a,⁎, Jeffrey W. Elder a,1, Lauren A. Baldwin a, Rachel W. Miller a, Chris P. DeSimone a, John R. van Nagell Jr a, Luis M. Samoyoa b, Dava S. West b, Emily V. Dressler c, Meng Liu c, Frederick R. Ueland a a b c
Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Kentucky, Markey Cancer Center, Lexington, KY, United States Department of Pathology, University of Kentucky, Markey Cancer Center, Lexington, KY, United States Division of Cancer Biostatistics, University of Kentucky, Markey Cancer Center, Lexington, KY, United States
a r t i c l e
i n f o
Article history: Received 15 November 2016 Received in revised form 30 January 2017 Accepted 31 January 2017 Available online xxxx Keywords: Endometrial cancer Intraoperative Surgical staging Lymph node Mayo Clinic algorithm, prospective
a b s t r a c t Objectives. Prospectively validate an intraoperative surgical staging algorithm to stratify patients with early endometrial cancer by risk of lymph node metastasis. Methods. Subjects with endometrial cancer clinically confined to the uterus were prospectively enrolled at an academic cancer center between Jan 2012 and Jun 2015. Study participants were stratified intraoperatively into two groups based on risk of nodal involvement using cell type, tumor grade, myometrial invasion, and tumor size in accordance with an established protocol from the Mayo Clinic. Low risk (LR) subjects received extrafascial hysterectomy with bilateral salpingo-oophorectomy; high risk (HR) patients received complete surgical staging including bilateral pelvic and para-aortic lymphadenectomy. Results. Of the 200 subjects enrolled, 194 were eligible for analysis. The algorithm identified 132 (68%) HR and 62 (32%) LR cancers. Of the HR subjects, 126 had lymphadenectomy performed with 14 (11%) positive for nodal metastases. Five HR subjects experienced disease recurrence. Of the 62 LR cancers, two patients developed disease recurrence. Ten LR cancers were upgraded to HR on final pathology due to lesion size (6) and grade (4). None of these patients experienced disease recurrence. The algorithm demonstrated 90% sensitivity (18/20) and 36% specificity (62/174) as determined by positive lymph nodes and/or disease recurrence. Conclusions. Intraoperative assessment of early endometrial cancer can be used to determine the extent of surgical staging. The studied algorithm has low specificity and modifications are necessary to better match the surgical procedure to the risk of metastatic cancer. Published by Elsevier Inc.
1. Introduction Endometrial cancer (EC) is the most common gynecologic malignancy in the United States, and the incidence continues to rise. In 2016, there are expected be 60,050 new cases of EC with an estimated 10,470 deaths [1]. The majority of women with EC present with early stage disease where survival rates exceed 80% following primary surgery [2]. In the United States, standard surgical staging for women with EC includes extrafascial hysterectomy, bilateral salpingo-oophorectomy, and lymphadenectomy. In 2014, the Society of Gynecologic Oncology Clinical Practice Endometrial Cancer Working Group
⁎ Corresponding author at: Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, University of Kentucky, Markey Cancer Center, 800 Rose Street, Lexington, KY 40536, United States. E-mail address: [email protected] (J.R. Lefringhouse). 1 Current affiliation: Division of Gynecologic Oncology, Greenville Hospital System Cancer Institute, Greenville, SC.
reported: 1) patients with grade 1–2 endometrioid tumors, b 50% myometrial invasion, and tumor of 2 cm or less are at low risk for recurrence and may not require a pelvic and para-aortic lymphadenectomy, and 2) lymphadenectomy may alter or eliminate the need for adjuvant therapy and its associated morbidity [3]. Today, there remains disagreement among specialists both on the necessity and extent of surgical staging for women with early EC [2,4–6]. The Mayo Clinic has published extensively on the surgical management of early EC, defining a subgroup of women at very low risk for lymph nodal involvement in which lymphadenectomy can be omitted. The authors developed an institutional algorithm that directs surgical treatment based on pathologic intraoperative consultation (IOC). Intraoperative stratification is based on the risk of nodal involvement: low risk (confined to corpus, grade 1 or 2 endometrioid type, myometrial invasion ≤50%, tumor diameter ≤ 2 cm), or high risk (does not meet low risk criteria). The low risk (LR) group is not required to undergo lymph node sampling or lymphadenectomy, while the high risk (HR) group receives bilateral pelvic and para-aortic lymphadenectomy [5,7]. In 2008,
http://dx.doi.org/10.1016/j.ygyno.2017.01.032 0090-8258/Published by Elsevier Inc.
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
2
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
the authors published results on 422 women treated with this algorithm. Of the 112 in the LR group not requiring lymphadenectomy, 22 (20%) had lymphadenectomy and all nodes were negative for metastases. In the HR group, 22% had metastatic nodal involvement identified by lymphadenectomy. The investigators determined that the five-year recurrence-free survival was 98% in low risk women and 54% in high risk patients (p b 0.001) [8]. Despite these encouraging reports, further study is needed to determine how the intraoperative algorithm performs outside the unique Mayo Clinic setting where the intraoperative pathology processing differs substantially from other institutions. This prospective investigation of women with early endometrial cancer was undertaken to evaluate the performance of the Mayo Clinic algorithm and its concordance with final pathology when using targeted intraoperative frozen section analysis. 2. Materials and methods Study participants who met eligibility criteria were consecutively enrolled and underwent surgical treatment at the University of Kentucky between January 2012 and June 2015. Institutional review board approval was obtained and the trial was registered on ClinicalTrials.gov (NCT01512810). Eligible participants had histologically confirmed EC clinically confined to the uterus. The use of preoperative imaging was not required but could be used at the discretion of the treating physician. Eligible subjects had a GOG performance status of 0, 1, or 2, and were surgical candidates for a complete hysterectomy, bilateral salpingo-oophorectomy, and pelvic and para-aortic lymphadenectomy. Subjects were included if a previous diagnosis of invasive cancer or chemotherapy was N5 years prior to enrollment; those with prior vaginal, pelvic or abdominal irradiation were excluded. The treating surgeon and patient collectively determined whether the operation was via laparotomy or a robotic-assisted laparoscopic approach. Laparoscopy alone was not offered because the participating surgeons prefer to perform lymphadenectomy with robotic-assistance. When surgically removed, the uterus and cervix were promptly sent to surgical pathology for intraoperative evaluation according to the IOC algorithm. The uterus was bivalved in the coronal axis. Visible endometrial lesions were measured in three dimensions with the longest dimension defined as the primary tumor diameter. At least one representative frozen section of the tumor was performed and a systematic intraoperative risk assessment was completed. If multiple lesions were noted, the largest lesion was measured and recorded. Histologic type was classified according to World Health Organization criteria. Architectural grading was based on degree of glandular differentiation according the FIGO guidelines and assignment of stage utilized the 2009 FIGO staging system [9]. In accordance with the Mayo Clinic criteria, the LR group was defined by the presence of all of the following criteria: endometrioid cell type with disease confined to the uterine corpus, any grade without myometrial invasion, grade 1 or 2 tumors with myometrial invasion less than or equal to 50%, and primary tumor diameter ≤ 2 cm. Low risk subjects underwent hysterectomy with bilateral salpingo-oophorectomy, but were not required to have pelvic and para-aortic lymphadenectomy. The remaining subjects were classified as HR and were surgically staged with bilateral pelvic (right and left external iliac, internal iliac, obturator, common iliac) and bilateral para-aortic lymphadenectomy to the level of the inferior mesenteric artery. The study employed a prospective non-randomized stratified observational design to estimate 2-year recurrence-free survival (RFS) rates. Sample size was determined a priori, using retrospective data from the University of Kentucky Endometrial Cancer Database; we predicted 2year RFS rates to be 92% in the low-risk and 82% in the high-risk subgroup. With the assumption that 60% of subjects would be categorized as low risk during IOC, 167 eligible patients would provide 80% power to estimate the 2-year RFS rate within ±5.8% for low risk and ±9.8% for high risk based on a two-sided exact 95% binomial confidence interval for a single proportion. If low-risk subgroup rates of 55% or 50% were observed,
the total sample size would be increased to 182 or 200 respectively to maintain similar precision. Thus, the final sample size was 200. Once subjects were classified as LR or HR, statistical analysis included Fisher's exact test and 2-sample t-tests to assess for differences between risk statuses for categorical and continuous variables respectively. Discordance between IOC and final path reports were calculated with corresponding Kappa statistics. Differences were considered statistically significant at p b 0.05. Statistical analyses were conducted using the SAS version 9.4 software package (SAS Institute, Inc.: Cary, NC). 3. Results A total of 200 subjects with EC were consecutively enrolled and 194 were eligible for analysis. Of the six ineligible subjects, three did not meet inclusion criteria, two declined surgery, and one had surgery at another facility (Fig. 1). The mean lymph node count for the 126 high risk patients who had lymph nodes removed was 15.4 (range 0–44). The mean lymph node count for the patients who had a completed pelvic and para-aortic lymphadenectomy was 17.7 (range 6–44). 3.1. Findings related to risk groups Based on intraoperative evaluation, 32% (62/194) were LR and 68% (132/194) were HR. The clinical and pathological characteristics are shown in Table 1. Five subjects in the LR group not requiring lymphadenectomy by protocol received at least partial lymphadenectomy. All lymph nodes were negative on final pathology, but one patient developed disease recurrence at 18 months after surgery. In all, two subjects (3%) in the LR group developed recurrent disease. They were the only patients in the LR group to receive additional therapy following surgery. Fourteen of 132 (10%) subjects in the HR group had positive lymph nodal metastases on final pathology. Intraoperatively, all were grade 2 or 3 tumors or had a lesion size N3 cm. Three patients had a lesion size of 2–3 cm. There were no lymph node metastases identified for grade 1 tumors with a lesion size ≤3 cm. Five of 14 (36%) had cervical involvement (Table 2). Of the 132 subjects in the HR group, 32% received either adjuvant radiation (11%), chemotherapy (14%), or both (7%). All HR subjects with lymph node metastases received adjuvant therapy. Of the seven subjects who developed disease recurrence, five were HR including one with positive lymph nodes and two with cervical involvement on final pathology. Two subjects with recurrence were in the LR cohort (Table 3). One LR subject was initially classified as HR by tumor size because the lesion grossly appeared to replace the entire surface of the endometrium. Lymph node dissection was initiated (seven negative para-aortic nodes), but a representative frozen section identified only complex atypical hyperplasia and the lymph node dissection was discontinued. The final pathology revealed a grade 1, endometrioid adenocarcinoma with a single focus of invasive disease confined to the endometrium. She developed a pelvic recurrence at two years and was treated with pelvic radiation; a supra-clavicular lymph node recurrence 10 months later was treated with chemotherapy; the patient is alive 29 months after completion of therapy. The second subject with a LR recurrence had a 1.5 cm, grade 1 endometrioid adenocarcinoma with 10% myometrial invasion on final pathology. A vaginal cuff recurrence at seven months was treated with whole pelvic radiation and vaginal cuff brachytherapy. The patient is alive and without definitive progression 26 months after her recurrence. During the mean follow up of 15.3 months (range 1–43), six subjects died: four were disease related and two were from causes unrelated to their cancer diagnosis. At this time, there is no difference in recurrencefree survival between the HR and LR groups on Kaplan-Meier analysis. 3.2. Findings related to individual algorithm components The discordance by risk group between IOC and final pathology was 5.7% (11/194, K = 0.861, p b 0.0001). Sixteen percent (10/62) of subjects
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
3
Fig. 1. CONSORT diagram summarizing study flow.
initially classified as LR were HR on final pathologic evaluation. Six were due to changes in size of lesion, all of which were grade 1 tumors and 4 were reclassified from endometrioid cell type intraoperatively to serous post-operatively. Disparities in myometrial invasion and grade were not responsible for any changes in risk group. None of the ten patients reclassified from LR to HR elected to undergo subsequent lymphadenectomy and none have recurred. The remaining patient (1/132) was intraoperatively considered HR but later deemed LR on final pathology based on re-measurement of tumor size. When considering each factor on univariate analysis, there was a high degree of concordance between IOC and final pathology for risk stratification (kappa = 0.861, p b 0.0001, as well as the individual risk components of tumor size categories (kappa = 0.8629), grade (kappa = 0.7078), histology (kappa = 0.6561) and myometrial invasion (kappa = 0.5628) (p b 0.0001 for each). Regarding myometrial invasion, 3.7% (6/164) of subjects with ‘none’ or ≤50% invasion were changed to N50%, while 17% (5/30) where changed from N50% to ≤50% invasion. The overall discrepancy rate for myometrial invasion when divided into these two categories (≤50% and N50%) was 5.7% (11/194). These discrepancies did not result in risk group reclassification. The discordance rate for IOC lesion size to final pathology was 4.3% (8/184). Of the 62 subjects with IOC tumor size ≤ 2 cm, six were upgraded to N2 cm, while two of 125 subjects with size N2 cm were downgraded. A similar discrepancy rate of 4.9% (9/184) would have
been observed if the cutoff for lesion size were changed to ‘any’ for grade 1 and b 3 cm for grade 2 tumors. For tumor grade, 5.7% (9/158) of grade 1–2 intraoperative tumors were grade 3 on final pathology, while 5.7% (2/35) grade 3 tumors were downgraded to grade 1–2 for a discrepancy rate of 5.7% (11/ 193). These grade changes did not result in a change of risk group. 4. Discussion There is ongoing debate over the utility and extent of lymphadenectomy for the surgical treatment of early stage EC. A number of key factors influence the discussion, including: the specifics of preoperative and intraoperative risk stratification, operative morbidity and mortality, cost of treatment, and patient survival. Even if patient survival is not affected by lymphadenectomy in women with early EC, tailoring the operation to match the extent of disease is an appealing concept to reduce the morbidity of treatment. The IOC algorithm is intended to segregate early endometrial cancers into LR and HR groups during surgery to give surgeons an intraoperative surrogate for factors associated with the risk of extra-uterine involvement on final pathology [10]. Twenty-five years ago, Schink and coauthors proposed tumor size as an important component of EC risk stratification [11]. More recently, depth of myometrial invasion, tumor grade and cell type have been added as significant factors [10–12]. Mariani and colleagues at the
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
4
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
Table 1 Patient demographics and disease related parameters by intraoperative risk for lymph node involvement. Intraoperative group
Number of patients Age at diagnosis (years) BMI Underweight + normal (b25) Overweight (25–30) Obese I (30–35) Obese II (35–40) Obese III (40+) Race White Black or African American Pre-op FIGO Grade, N (%) Grade 1 Grade 2 Grade 3 Grade unknown, not stated Pre-op histologic subtype Adenocarcinoma, NOS Endometrioid Serous Clear cell Adenosquamous Carcinosarcoma Undifferentiated Not applicable Lymphadenectomy Yes No Type of hysterectomy Total abdominal Robot-assisted laparoscopic
High risk
Low risk
p-Value
132 61.4 (10.0)
62 56.3 (9.4)
0.0009
17 (13.0%) 23 (17.6%) 34 (26.0%) 24 (18.3%) 33 (25.2%)
11 (18.0%) 7 (11.5%) 12 (19.7%) 13 (21.3%) 18 (29.5%)
0.5772
128 (97.0%) 4 (3.0%)
62 (100.0%) 0 (0.0%)
0.3082
66 (50.0%) 33 (25.0%) 30 (22.7%) 3 (2.3%)
47 (75.81%) 13 (21.0%) 2 (3.2%) 0 (0.0%)
9 (6.8%) 109 (82.6%) 8 (6.1%) 2 (1.5%) 1(0.8%) 1(0.8%) 1(0.8%) 1(0.8%)
4 (6.4%) 58(93.6%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%)
126 (95.5%) 6 (4.6%)
5 (8.1%) 57 (91.9%)
96 (62.5%) 36 (37.5%)
44 (59.0%) 18 (40.9%)
0.0003
0.4394
b0.0001
0.7987
p-values were calculated from t-test for continuous variables and chi-squared test and Fisher's exact test for categorical variables; NOS - not otherwise specified.
Mayo Clinic combined these features into a specific intraoperative algorithm to be used in the surgical triage of early endometrial cancers. They published a 5% rate of lymph node involvement and a 96% recurrencefree survival in the LR group [7]. Other investigators have studied the Mayo Clinic algorithm. Convery and coauthors completed a two center, retrospective chart review of 110 subjects with early EC and reported that 2.4% of women with LR cancers developed disease recurrence [13]. Vargas and co-investigators utilized Surveillance, Epidemiology, and End Results (SEER) data from 19,329 women to confirm that the LR group by Mayo criteria had a 1.4% risk of lymph node metastasis
compared to 6.4% with HR features [14]. In a community setting, Bell and coauthors retrospectively evaluated the outcomes of 179 patients found to be LR by intraoperative assessment and surgically managed without lymphadenectomy and reported a 1.7% recurrence risk and five year overall survival of 95.8% [15]. The findings of our prospective investigation confirm that the Mayo Clinic algorithm can be successfully employed at other institutions. In agreement with previous reports, we find a 3% (2/62) recurrence rate for the LR group, one distant (pelvic sidewall and supra-clavicular node) and one local (vaginal cuff). The most challenging aspect of implementing the Mayo Clinic algorithm is standardizing the processing of surgical pathology specimens. In brief, the Mayo clinic has a robust, intraoperative processing capability where multiple frozen sections are performed (often the entire specimen) and the final pathologic diagnosis is rendered immediately. In rare instances, the final diagnosis is delayed for further processing with immunohistochemical staining. This practice is not uniform in the United States. At the other extreme, facilities without frozen section analysis have attempted to circumvent this obstacle by combining the preoperative biopsy with intraoperative tumor dimensions [13]. Evaluating our specimen processing results, the risk category differed from IOC to final pathology in 5.7% of patients analyzed. Ten of the 62 (16%) LR subjects were upgraded, six for lesion size and four for grade (all endometrioid to grade 3 serous cell type). Three with serous histology on final pathology received adjuvant chemotherapy with carboplatin and paclitaxel. None of the ten received lymphadenectomy and none have recurred at a mean follow up of 15.3 months. Only one subject was downgraded on final pathology for tumor size. Thus, the IOC with targeted frozen section analysis compared favorably with final pathology in terms of individual risk factors and relevant clinical outcomes. To further minimize discordance, the authors recommend that the IOC include at least one frozen section (preferably two or more) with specific attention to intraoperative grade, cell type, depth of myometrial invasion, and cervical involvement. The IOC frozen section should confirm that the measured lesion is malignant and reassess tumor grade. Tumor grade should not be lowered even if it differs from the preoperative result, and a cancer diagnosis should not be overturned from a preoperative biopsy. Whenever possible, preoperative biopsies should be reviewed at the treating institution prior to surgery. The intraoperative algorithm demonstrated 90% sensitivity (18/20) and 36% specificity (62/174) for positive lymph nodes and disease recurrence. The algorithm performed well at identifying EC at HR of extra-uterine disease; however, by recommending lymphadenectomy for over two-thirds of the study population, the IOC criteria mis-
Table 2 Pathologic findings and outcomes for subjects with positive lymph nodes. Patient #
1 2 3 4 5 6 7 8 9 10 11 12 13a 14
Intraoperative pathology
Final pathology
Histology
Tumor size (cm)
Tumor grade
MI
Histology
Tumor size (cm)
Tumor grade
MI
Cervix involved
Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Serous
N3 N3 (2–3) N3 N3 N3 N3 N3 N3 (2–3) N3 (2–3) N3 N3
1 1 2 1 1 3 1 1 3 2 3 3 2 3
b50% N50% b50% b50% N50% b50% b50% N50% b50% b50% b50% b50% b50% b50%
Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid Serous Endometrioid
N3 N3 (2–3) N3 N3 N3 N3 N3 N3 (2–3) N3 (2–3) N3 N3
1 2 2 3 2 3 2 2 3 1 2 3 3 3
b50% N50% b50% N50% b50% N50% b50% N50% b50% b50% b50% b50% b50% b50%
− + + − − + − + − + − − − −
All patients with positive lymph nodes were intraoperatively high risk per Intraoperative algorithm and post-operative pathology. MI - myometrial invasion. a One patient had both +LN and + disease recurrence.
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032
J.R. Lefringhouse et al. / Gynecologic Oncology xxx (2017) xxx–xxx
5
Table 3 Pathologic findings and outcomes for subjects with disease recurrence. Patient #
Intraoperative evaluation Risk group
Histology
Tumor size (cm)
Tumor grade
Lymph-adenectomy Risk group
Histology
Tumor size (cm)
Tumor grade
Lymph node positive
Cervix involved
1 2 3 4 5 6 7a
High High High Low Low High High
Endometrioid Undifferentiated Adenosquamous Endometrioid Endometrioid Endometrioid Endometrioid
N3 N3 N3 ≤2 ≤2 (2–3) N3
2 3 3 1 1 3 2
Yes Yes Yes Yes No Yes Yes
Endometrioid Endometrioid Non-Endometrioid Endometrioid Endometrioid Endometrioid Serous
N3 N3 N3 ≤2 ≤2 (2–3) N3
3 3 3 1 1 3 3
− − − − N/A − +
+ − − − − + −
a
Final evaluation
High High High Low Low High High
One patient had both +LN and + disease recurrence.
categorized many LR cases as HR. In a recent SEER review, Vargas published similar results of the Mayo criteria with a test sensitivity of 85% (337/395) and specificity of 28% (4039/13,969). Similar to Mayo Clinic publications, Vargas reported that only 21% of subjects were classified as LR. To increase algorithm specificity, they proposed that the criteria be modified to selectively reduce the number of subjects included in the HR group while limiting an increase the number of false negative results [14]. The current study confirms that the Mayo criteria can be successfully implemented at other centers, and that published population-based reports are consistent with the findings of this prospective investigation. Our study demonstrated lymph node involvement in the LR group of 1.8%, 0%, and 0% compared to SEER data of 0.8%, 2.1% and 0% for grade 1, 2 and 3 tumors, respectively. Given that most cancers are categorized as high risk by the Mayo system, Vargas et al. suggested that the LR group be expanded to include grade 1 endometrioid tumors with b50% myometrial invasion regardless of lesion size, grade 2 tumors with lesion size b 3 cm, and grade 3 tumors with no myometrial invasion. In the Vargas SEER review, the LR group using this modified algorithm would have had a positive lymph node rate of 2.3%, 2.0% and 0% for grade 1, 2 and 3 tumors, respectively. This new algorithm would have a calculated test sensitivity of 63% and specificity of 66%. Extrapolating with our prospective data, the Vargas-proposed algorithm for the LR group would theoretically result in a positive lymph node rate of 2.2%, 0% and 0%, respectively. This change would increase specificity to 65%, decrease sensitivity to 75%, and decrease the number of women receiving pelvic and para-aortic lymphadenectomy from 68% to 38%. However, this change would potentially increase the number of false negative results from two to five. The strengths of this investigation are its prospective design, centralized pathology review, the consistent implementation of the IOC algorithm at a single institution, and both treatment and follow-up were performed by the same physicians. Our strategy of representative frozen section analysis during IOC is likely to be available at all medical centers with practicing gynecologic oncologists. A limitation of the study is the single institution design which limits generalizability. A second weakness is the short follow-up time of 15.3 months which excludes a meaningful survival analysis. The findings of this investigation indicate that the process of selective surgical staging with targeted frozen section for women with early endometrial cancer is reproducible and effective at identifying HR cancers with 90% sensitivity. There are two primary benefits to stratifying patients based on their risk of nodal metastases during surgery: 1) omission of lymphadenectomy in the LR group, and 2) tailoring adjuvant treatment for the HR group, possibly avoiding radiation when lymph nodes are histologically negative [15]. Future multi-institutional, prospective trials will determine if this strategy can improve the cost and safety of surgical treatment, without diminishing survival
outcomes. Since two-thirds of early EC are classified as high risk by the current IOC algorithm, we see an opportunity to thoughtfully expand the LR group to achieve a better clinical balance of surgical risk and treatment outcomes. Conflict of interest statement Emily Dressler PhD received grant funding through the University of Kentucky Markey Cancer Center Biostatistics Shared Resource Facility during the conduct of this study. NIH (NCI) P30 CA177558
References [1] R.L. Siegel, K.D. Miller, A. Jemal, Cancer statistics, 2016, CA Cancer J. Clin. 66 (1) (2016) 7–30. [2] H.C. Kitchener, E.L. Trimble, Intergroup ftECWGotGC. Endometrial cancer state of the science meeting, Int. J. Gynecol. Cancer 19 (1) (2009) 134–140. [3] William M. Burke JO, Mario Leitao, Emery Salom, Paola Gehrig, Alexander B. Olawaiye, Molly Brewer, Dave Boruta, Jeanine Villella, Tom Herzog, Fadi Abu Shahin, Endometrial cancer: a review and current management strategies: part I, Gynecol. Oncol. 134 (2) (2014) 8. [4] H. Kitchener, A. Swart, Q. Qian, C. Amos, M. Parmar, Efficacy of systematic pelvic lymphadenectomy in endometrial cancer (MRC ASTEC trial): a randomised study, Lancet 373 (9658) (2009) 125–136. [5] S.C. Dowdy, B.J. Borah, J.N. Bakkum-Gamez, A.L. Weaver, M.E. McGree, L.R. Haas, et al., Prospective assessment of survival, morbidity, and cost associated with lymphadenectomy in low-risk endometrial cancer, Gynecol. Oncol. 127 (1) (2012) 5–10. [6] P.B. Panici, S. Basile, F. Maneschi, A.A. Lissoni, M. Signorelli, G. Scambia, et al., Systematic pelvic lymphadenectomy vs no lymphadenectomy in early-stage endometrial carcinoma: randomized clinical trial, J. Natl. Cancer Inst. 100 (23) (2008) 1707–1716. [7] A. Mariani, M.J. Webb, G.L. Keeney, M.G. Haddock, G. Calori, K.C. Podratz, Low-risk corpus cancer: is lymphadenectomy or radiotherapy necessary? Am. J. Obstet. Gynecol. 182 (6) (2000) 1506–1519. [8] A. Mariani, S.C. Dowdy, W.A. Cliby, B.S. Gostout, M.B. Jones, Wilson TO, et al., Prospective assessment of lymphatic dissemination in endometrial cancer: a paradigm shift in surgical staging, Gynecol. Oncol. 109 (1) (2008) 11–18. [9] FIGO, Staging for carcinoma of the vulva, cervix, and corpus uteri, Int. J. Gynaecol. Obstet. 125 (2) (2014) 97–98. [10] H.M. Keys, J.A. Roberts, V.L. Brunetto, R.J. Zaino, N.M. Spirtos, J.D. Bloss, et al., A phase III trial of surgery with or without adjunctive external pelvic radiation therapy in intermediate risk endometrial adenocarcinoma: a gynecologic oncology group study, Gynecol. Oncol. 92 (3) (2004) 744–751. [11] J.C. Schink, A.W. Rademaker, D. Miller, J.R. Lurain, Tumor size in endometrial cancer, Cancer 67 (11) (1991) 2791–2794. [12] W.T. Creasman, C.P. Morrow, B. Bundy, H.D. Homesley, J. Graham, P. Heller, Surgical pathologic spread patterns of endometrial cancer. A gynecologic oncology group study, Cancer 60 (8 Suppl) (1987) 2035–2041. [13] M.M. AlHilli, A. Mariani, Preoperative selection of endometrial cancer patients at low risk for lymph node metastases: useful criteria for enrollment in clinical trials, J. Gynecol. Oncol. 25 (4) (2014) 267–269. [14] R. Vargas, J.A. Rauh-Hain, J. Clemmer, R.M. Clark, A. Goodman, W.B. Growdon, et al., Tumor size, depth of invasion, and histologic grade as prognostic factors of lymph node involvement in endometrial cancer: a SEER analysis, Gynecol. Oncol. 133 (2) (2014) 216–220. [15] C. Sharma, I. Deutsch, S.N. Lewin, W.M. Burke, Y. Qiao, X. Sun, et al., Lymphadenectomy influences the utilization of adjuvant radiation treatment for endometrial cancer, Am. J. Obstet. Gynecol. 205 (6) (2011) 562 (e1-.e9).
Please cite this article as: J.R. Lefringhouse, et al., Prospective validation of an intraoperative algorithm to guide surgical staging in early endometrial cancer, Gynecol Oncol (2017), http://dx.doi.org/10.1016/j.ygyno.2017.01.032