- Email: [email protected]
Contribution of the frozen section in the management on endometrial cancer
Abstracts / Gynecologic Oncology 125 (2012) S3–S167
architecture and presence or absence of microcystic elongated and fragmented glands (MELF), single cell invasion, lymphovascular invasion (LVI), lower uterine segment (LUS) and cervix (CX) involvement, and numbers of pelvic (PLN) and para-aortic (PALN) LNs sampled. Results: Three hundred two cases were reviewed: LN+or ED+, 96; LN-/ ED-, 206. Patients' ages ranged from 23 to 91 yrs (median 61). The following histopathologic variables were noted for the LN+or ED+ group: tumor size ≥2 cm 93/96 (97%), MI N50% 54/96 (56%), MELF 61/96 (64%), single cell invasion 28/96 (29%), LVI 77/96 (80%), N20% solid 31/96 (32%), papillary architecture present 58/96 (60%), LUS involved 64/96 (67%), and CX involved 40/96 (42%). For the LN-/ED- group, the results were as follows: tumor size ≥2.0 cm 149/206 (72%), MI N50% 56/206 (27%), MELF 68/206 (33%), single cell invasion 16/206 (8%), LVI 52/206 (25%), N20% solid 44/206 (21%), papillary architecture present 100/206 (49%), LUS involved 75/206 (36%), and CX involved 23/206 (11%). The Table provides univariate logistic regression results modeling LNM or ED. There was no evidence of a difference in the number of pelvic or paraaortic LNs sampled between groups (p=0.10 and 0.64, respectively). Conclusions: Tumor size ≥2 cm, % papillary architecture, MELF, single cell invasion, LVI, LUS and CX involvement, MI N50%, and number of PALNs sampled were found to be significant predictors of the odds of LNM or ED. Presence of papillary architecture and % solid component N20% also suggested increased odds (p= 0.0658 and 0.0527, respectively). Age, FIGO grade 2 vs 1, and number of PLNs sampled were not significant predictors. This multi-institutional study validates previous findings of MI and LV as predictors of LNM and ED. Additionally, MELF and single cell invasion patterns are also shown to be predictors of LNM or ED. Attention to the pattern of invasion during frozen section may identify patients who could benefit from LND. Further study is required to determine a multivariate model to predict advanced stage and to study whether these features could also be predictors of recurrence. Table Univariate Logistic Regression Results Modeling LNM or ED Variable
Odds Ratio
Confidence Interval
p-value
Tumor size ≥2 cm % solid N 20% Papillary architecture % Papillary architecture* MELF Single cell invasion LVI LUS involved Cx involved MI N 50% # PALN sampled
13.95 1.72 1.59 1.02 3.54 4.58 12.00 3.31 5.93 3.44 1.09
3.31-58.89 0.99-2.93 0.97-2.60 1.01-1.03 2.13-5.87 2.36-8.89 6.64-21.7 1.91-5.75 3.28-10.73 2.08-5.72 1.03-1.15
0.0003 0.0527 0.0658 0.0058 b 0.0001 b 0.0001 b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.0048
*Several cutoffs were statistically significant with the lowest p-value at 0.0001 for N30% papillary architecture.
doi:10.1016/j.ygyno.2011.12.381
381 Unraveling the cause of death in endometrial cancer: A study of 2513 patients from a single institution J. Barlin, W. Wysham, A. Ferda, F. Khoury Collado, D. Cassella, K. Alektiar, M. Hensley, D. Chi, R. Barakat, N. Abu-Rustum. Memorial Sloan-Kettering Cancer Center, New York, NY. Objective: The cause of death from endometrial cancer is poorly described, whereas it is well accepted that ovarian cancer patients (pts) mainly die from malignant bowel obstruction and cervical cancer pts historically die from uremia. The purpose of this study was to describe the location of disease at time of death for endometrial cancer pts who died of their disease. Methods: Excluding leiomyosarcoma, endometrial stromal sarcoma, and adenosarcoma, all pts diagnosed with endometrial cancer from
S155
1/1993 through 12/2010 were included. Histology was separated into 2 groups: endometrioid and high-risk (HR) histologies [papillary serous (PS), clear cell (CC), carcinosarcoma (CS), and undifferentiated (UN)]. Status was defined as no evidence of disease (NED), alive with disease (AWD), and dead. Pts who died were divided into dead of disease (DOD), dead of other (DOO), and dead lost to follow-up (DLTF). Patterns of disease spread at death were documented from most recent exam and imaging studies and were categorized into 3 groups: pelvic only, abdominal with or without pelvic, and distant. Results: During the study period, 2513 pts met inclusion criteria. Median age at diagnosis was 62. Histology was: endometrioid 1949 (78%), PS 314 (13%), CC 77 (3%), CS 151(6%), UN 22 (1%). 1988 FIGO stage was: stage I 1763 (70%), stage II 145 (6%), stage III 416 (17%), and stage IV 189 (8%). At the time of this study, 1867 (74%) were NED, 232 (9%) were AWD, and 414 (16%) were dead. Of the 16% dead, 224 (54%) were DOD, 84 (20%) were DOO, and 106 (26%) were DLTF. Using chi-square, there was no significant difference in pattern of location of disease between endometrioid and HR histologies (p = 0.36). Of the 224 pts DOD, the location of disease at time of death insert: (Table) was: pelvic 23 (10%), abdominal 83 (37%), and distant 118 (53%). 29 (35%) of the 83 patients with abdominal disease had liver involvement, and 92 (78%) of 118 pts with distant disease had lung metastases. 90% of pts who died of their endometrial cancer had disease outside of the pelvis at the time of death. Conclusions: These data suggest that death from endometrial cancer is largely due to abdominal (liver) and distant (lung) metastases, and this pattern of disease appears similar in endometrioid and HR histologies. Although most pts diagnosed with endometrial cancer will be cured, the vast majority of the pts who died of their disease had metastases beyond the pelvis at the time of death.
Table Location of disease at time of death for endometrial cancer patients who died of disease. Dead of Disease
Endometrioid Histology
High-risk Histology
Total
Pelvic Abdominal Distant Total patients
11 (12.5%) 28 (31.8%) 49 (55.7%) 88
12 (8.8%) 55 (40.4%) 69 (50.7%) 136
23 (10.3%) 83 (37.1%) 118 (52.7%) 224
doi:10.1016/j.ygyno.2011.12.382
382 Contribution of the frozen section in the management on endometrial cancer N. Bourdel1, A. Spirtos2, M. Fairbarn2, V. Pustavoitava2, A. Tergas2, R. Giuntoli2. 1Kelly Gynecologic Oncology Service, Johns Hopkins Medical Institutions, Baltimore, MD, 2Johns Hopkins Kimmel Cancer Center, Baltimore, MD. Objective: To assess accuracy of frozen section evaluation (FSE) in determining whether lymphadenectomy is indicated in the surgical evaluation of patients with endometrial cancer as compared to intraoperative gross evaluation (GE) and to no intraoperative evaluation (NE). Methods: We performed a retrospective study of all women undergoing hysterectomy for endometrial cancer at the Kelly Gynecologic Oncology Service (Johns Hopkins Hospital, Baltimore, MD) between January 1989 and December 2007. Theoretical indications for lymphadenectomy were based on criteria described by Mariani et al. Lymphadenectomy was considered not indicated for low-risk patients grade 1 or 2 endometrioid histology, myometrial invasion
S156
Abstracts / Gynecologic Oncology 125 (2012) S3–S167
(MI) ≤50%, and tumor size ≤2 cm. We separated patients into 3 groups: final pathology only (NE: no intraoeprative evaluation, with the assumption that lymph node dissection should have been performed for every cases), gross evaluation (GE), and frozen section evaluation (FSE). Comparing theoretical intraoperative assessment for lymphadenectomy to actual need for nodes, sensitivity, specificity, PPV, and NPV were calculated. Results: Six hundred and twenty-four patients were included: 327 in the FSE group, 180 in the GE group, and 117 in the NE group. The 3 groups did not differ significantly by age, grade, stage, histology, BMI, or parity. The sensitivity was 100, 81.8, and 94.9 and the specificity was 0, 90, and 99.4%, respectively for the NE, GE, and FSE groups. The positive predictive value (PPV) was 55.6%, 92.8%, and 99.4% and the negative predictive value was N/A, 75.9%, and 94.3% (p b 0.05), respectively for the NE, GE, and FSE groups. The kappa coefficient for GE was 0.69, indicating moderate-to-substantial agreement; the kappa coefficient for FSE was 0.93, indicating almost perfect agreement. In the NE group, 44.4% (n = 52) had an unindicated lymphadenectomy compared to 3.9% (n = 7) in the GE group and 0.3 % (n = 1) in the FSE group. 11.1% (n = 20) in the GE group and 2.7% (n = 9) in the FSE group were upstaged based on final pathology Conclusions: In this series, the specificity, sensitivity, and PPV of the FSE were very high (99%, 95%, and 99.4%, respectively). The addition of FSE to the risk-predicting diagnostic procedure decreased the rate of unindicated lymphadenectomy by 44% as compared to the NE group and by a factor 10 compared to GE only, with a decrease in the need for a second restaging surgery. FSE is an essential component of the surgical risk-predicting diagnostic procedure. doi:10.1016/j.ygyno.2011.12.383
383 Lymphovascular space invasion is an isolated poor prognostic factor for recurrence among women with intermediate to high-risk early stage endometrioid endometrial adenocarcinoma of the uterus L. Weinberg1, C. Kunos2, K. Zanotti2. 1The Cleveland Clinic Foundation, Cleveland, OH, 2Case Western Reserve - MacDonald Women's Hospital, Cleveland, OH. Objective: While previous studies have also shown lymphovascular space invasion (LVSI) to increase risk for recurrent endometrioid uterine cancer (EC), its significance relative to other known risk factors has not been defined. As such, there is some uncertainty regarding whether this finding should independently impact adjuvant treatment decisions in women with early-stage disease. Our aim was to study prognostic factors for recurrence in women with earlystage intermediate-to high-risk EC. Methods: Women with surgically treated stage I or II EC from 1994 to 2010 were included in this study if they had LVSI, FIGO grade 2 or 3 histology (G2/3), or outer half myometrial invasion (OI). Patients were excluded if their follow-up was less than 6 months. Demographic, medical, and oncologic information was obtained from chart review. We performed univariate and multivariate logistic regression analyses and Fisher's exact test for 2-way analyses of categorical variables to identify prognostic factors for vaginal, pelvic, and distant recurrences. Results: Three hundred forty-eight patients were identified, 3% were excluded. The remaining 338 patients had a median age of 68 years, a median BMI of 30.7 and a median follow-up of 60 months (range 6– 194). Seventy-three percent of patients had a lymphadenectomy. Thirty percent of patients had LVSI, 7% had cervical stromal invasion, 84% had G2/3, and 43% had OI. Adjuvant treatments included vaginal brachytherapy (61%), pelvic radiotherapy (28%), and systemic therapy (6%). The
rate of recurrence was: overall 17.2%, pelvic 10.9%, vaginal cuff 8.0%, and distant 10.7%. In univariate analyses, only LVSI and depth of myometrial invasion were significantly associated with overall recurrence. However in multivariate analysis after adjusting for age, BMI, grade, depth of invasion, cervical invasion, lymphadenectomy, and adjuvant treatment (s), only LVSI was a significant independent risk factor for total recurrence (OR = 3.07, 95% CI 1.63 -5.83). Similarly, LVSI was the only significant risk factor for all types of recurrence (pelvic, distant, and vaginal cuff; p b 0.01). Conclusions: Our study found LVSI to be an independent prognostic factor for both pelvic and distant recurrences and a better predictor than all other risk factors that define the intermediate-to high-risk category of early-stage EC. Adjuvant therapies should be considered in patients with LVSI, even in the absence of other high-risk features. doi:10.1016/j.ygyno.2011.12.384
384 The importance of applying a sentinel lymph node mapping algorithm in endometrial cancer staging: Beyond removal of blue nodes J. Barlin, F. Khoury Collado, C. Kim, D. Cassella, D. Chi, Y. Sonoda, K. Alektiar, D. DeLair, R. Barakat, N. Abu-Rustum. Memorial Sloan-Kettering Cancer Center, New York, NY. Objective: To determine the effectiveness of a surgical sentinel lymph node (SLN) mapping algorithm in detecting metastatic endometrial cancer while minimizing the need for complete lymphadenectomy (LND). Methods: A prospective database of all patients who underwent lymphatic mapping for endometrial cancer was reviewed. Cervical injection of blue dye was used in all cases. The surgical algorithm (Figure) is as follows: 1. Peritoneal and serosal evaluation and washings; 2. Retroperitoneal evaluation, including excision of all mapped SLNs and any suspicious node regardless of mapping; 3. If there is no mapping on a hemi-pelvis, a side-specific pelvic, common iliac, and interiliac LND is performed. Para-aortic LND is at attending's discretion. We retrospectively applied the algorithm to evaluate its performance. Results: Between 9/2005 and 4/2011, 498 patients received a cervical injection of blue dye for SLN mapping. At least 1 lymph node was removed in 95% of cases (474/498), and at least 1 SLN was identified in 81% of cases (401/498). There was unilateral pelvic mapping in 29% (146/498) and bilateral pelvic mapping in 51% (253/498) of cases. Over 95% (384/401) of mapped SLNs were limited to the pelvis, 4% (15/401) were in both pelvic and para-aortic regions, and there was para-aortic-only mapping in 0.5% (2/401). Median SLN count was 3 (range 1–15), and median total lymph node count was 10 (range 1–55). SLN correctly diagnosed 40 of 47 patients with nodal metastases who had at least 1 SLN mapped, resulting in a falsenegative rate of 15%. After applying the algorithm, the false negative rate for detecting nodal metastasis dropped to 2%. Only 1 patient whose lymph node spread would not have been caught by the algorithm had an isolated positive right para-aortic lymph node with a negative ipsilateral SLN and pelvic LND. Conclusions: Satisfactory SLN mapping in endometrial cancer requires adherence to a surgical SLN algorithm and goes beyond just the removal of blue SLNs. Removal of any suspicious nodes along with side-specific lymphadenectomy for failed mapping are an integral part of this algorithm. The proposed algorithm may serve as a potential middle ground in the controversy of endometrial cancer surgical staging by providing a reasonably low false, negative rate, while sparing complete bilateral LND in the majority of cases. Isolated para-aortic lymphatic spread remains a known but small risk in the majority of patients.
architecture and presence or absence of microcystic elongated and fragmented glands (MELF), single cell invasion, lymphovascular invasion (LVI), lower uterine segment (LUS) and cervix (CX) involvement, and numbers of pelvic (PLN) and para-aortic (PALN) LNs sampled. Results: Three hundred two cases were reviewed: LN+or ED+, 96; LN-/ ED-, 206. Patients' ages ranged from 23 to 91 yrs (median 61). The following histopathologic variables were noted for the LN+or ED+ group: tumor size ≥2 cm 93/96 (97%), MI N50% 54/96 (56%), MELF 61/96 (64%), single cell invasion 28/96 (29%), LVI 77/96 (80%), N20% solid 31/96 (32%), papillary architecture present 58/96 (60%), LUS involved 64/96 (67%), and CX involved 40/96 (42%). For the LN-/ED- group, the results were as follows: tumor size ≥2.0 cm 149/206 (72%), MI N50% 56/206 (27%), MELF 68/206 (33%), single cell invasion 16/206 (8%), LVI 52/206 (25%), N20% solid 44/206 (21%), papillary architecture present 100/206 (49%), LUS involved 75/206 (36%), and CX involved 23/206 (11%). The Table provides univariate logistic regression results modeling LNM or ED. There was no evidence of a difference in the number of pelvic or paraaortic LNs sampled between groups (p=0.10 and 0.64, respectively). Conclusions: Tumor size ≥2 cm, % papillary architecture, MELF, single cell invasion, LVI, LUS and CX involvement, MI N50%, and number of PALNs sampled were found to be significant predictors of the odds of LNM or ED. Presence of papillary architecture and % solid component N20% also suggested increased odds (p= 0.0658 and 0.0527, respectively). Age, FIGO grade 2 vs 1, and number of PLNs sampled were not significant predictors. This multi-institutional study validates previous findings of MI and LV as predictors of LNM and ED. Additionally, MELF and single cell invasion patterns are also shown to be predictors of LNM or ED. Attention to the pattern of invasion during frozen section may identify patients who could benefit from LND. Further study is required to determine a multivariate model to predict advanced stage and to study whether these features could also be predictors of recurrence. Table Univariate Logistic Regression Results Modeling LNM or ED Variable
Odds Ratio
Confidence Interval
p-value
Tumor size ≥2 cm % solid N 20% Papillary architecture % Papillary architecture* MELF Single cell invasion LVI LUS involved Cx involved MI N 50% # PALN sampled
13.95 1.72 1.59 1.02 3.54 4.58 12.00 3.31 5.93 3.44 1.09
3.31-58.89 0.99-2.93 0.97-2.60 1.01-1.03 2.13-5.87 2.36-8.89 6.64-21.7 1.91-5.75 3.28-10.73 2.08-5.72 1.03-1.15
0.0003 0.0527 0.0658 0.0058 b 0.0001 b 0.0001 b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.0048
*Several cutoffs were statistically significant with the lowest p-value at 0.0001 for N30% papillary architecture.
doi:10.1016/j.ygyno.2011.12.381
381 Unraveling the cause of death in endometrial cancer: A study of 2513 patients from a single institution J. Barlin, W. Wysham, A. Ferda, F. Khoury Collado, D. Cassella, K. Alektiar, M. Hensley, D. Chi, R. Barakat, N. Abu-Rustum. Memorial Sloan-Kettering Cancer Center, New York, NY. Objective: The cause of death from endometrial cancer is poorly described, whereas it is well accepted that ovarian cancer patients (pts) mainly die from malignant bowel obstruction and cervical cancer pts historically die from uremia. The purpose of this study was to describe the location of disease at time of death for endometrial cancer pts who died of their disease. Methods: Excluding leiomyosarcoma, endometrial stromal sarcoma, and adenosarcoma, all pts diagnosed with endometrial cancer from
S155
1/1993 through 12/2010 were included. Histology was separated into 2 groups: endometrioid and high-risk (HR) histologies [papillary serous (PS), clear cell (CC), carcinosarcoma (CS), and undifferentiated (UN)]. Status was defined as no evidence of disease (NED), alive with disease (AWD), and dead. Pts who died were divided into dead of disease (DOD), dead of other (DOO), and dead lost to follow-up (DLTF). Patterns of disease spread at death were documented from most recent exam and imaging studies and were categorized into 3 groups: pelvic only, abdominal with or without pelvic, and distant. Results: During the study period, 2513 pts met inclusion criteria. Median age at diagnosis was 62. Histology was: endometrioid 1949 (78%), PS 314 (13%), CC 77 (3%), CS 151(6%), UN 22 (1%). 1988 FIGO stage was: stage I 1763 (70%), stage II 145 (6%), stage III 416 (17%), and stage IV 189 (8%). At the time of this study, 1867 (74%) were NED, 232 (9%) were AWD, and 414 (16%) were dead. Of the 16% dead, 224 (54%) were DOD, 84 (20%) were DOO, and 106 (26%) were DLTF. Using chi-square, there was no significant difference in pattern of location of disease between endometrioid and HR histologies (p = 0.36). Of the 224 pts DOD, the location of disease at time of death insert: (Table) was: pelvic 23 (10%), abdominal 83 (37%), and distant 118 (53%). 29 (35%) of the 83 patients with abdominal disease had liver involvement, and 92 (78%) of 118 pts with distant disease had lung metastases. 90% of pts who died of their endometrial cancer had disease outside of the pelvis at the time of death. Conclusions: These data suggest that death from endometrial cancer is largely due to abdominal (liver) and distant (lung) metastases, and this pattern of disease appears similar in endometrioid and HR histologies. Although most pts diagnosed with endometrial cancer will be cured, the vast majority of the pts who died of their disease had metastases beyond the pelvis at the time of death.
Table Location of disease at time of death for endometrial cancer patients who died of disease. Dead of Disease
Endometrioid Histology
High-risk Histology
Total
Pelvic Abdominal Distant Total patients
11 (12.5%) 28 (31.8%) 49 (55.7%) 88
12 (8.8%) 55 (40.4%) 69 (50.7%) 136
23 (10.3%) 83 (37.1%) 118 (52.7%) 224
doi:10.1016/j.ygyno.2011.12.382
382 Contribution of the frozen section in the management on endometrial cancer N. Bourdel1, A. Spirtos2, M. Fairbarn2, V. Pustavoitava2, A. Tergas2, R. Giuntoli2. 1Kelly Gynecologic Oncology Service, Johns Hopkins Medical Institutions, Baltimore, MD, 2Johns Hopkins Kimmel Cancer Center, Baltimore, MD. Objective: To assess accuracy of frozen section evaluation (FSE) in determining whether lymphadenectomy is indicated in the surgical evaluation of patients with endometrial cancer as compared to intraoperative gross evaluation (GE) and to no intraoperative evaluation (NE). Methods: We performed a retrospective study of all women undergoing hysterectomy for endometrial cancer at the Kelly Gynecologic Oncology Service (Johns Hopkins Hospital, Baltimore, MD) between January 1989 and December 2007. Theoretical indications for lymphadenectomy were based on criteria described by Mariani et al. Lymphadenectomy was considered not indicated for low-risk patients grade 1 or 2 endometrioid histology, myometrial invasion
S156
Abstracts / Gynecologic Oncology 125 (2012) S3–S167
(MI) ≤50%, and tumor size ≤2 cm. We separated patients into 3 groups: final pathology only (NE: no intraoeprative evaluation, with the assumption that lymph node dissection should have been performed for every cases), gross evaluation (GE), and frozen section evaluation (FSE). Comparing theoretical intraoperative assessment for lymphadenectomy to actual need for nodes, sensitivity, specificity, PPV, and NPV were calculated. Results: Six hundred and twenty-four patients were included: 327 in the FSE group, 180 in the GE group, and 117 in the NE group. The 3 groups did not differ significantly by age, grade, stage, histology, BMI, or parity. The sensitivity was 100, 81.8, and 94.9 and the specificity was 0, 90, and 99.4%, respectively for the NE, GE, and FSE groups. The positive predictive value (PPV) was 55.6%, 92.8%, and 99.4% and the negative predictive value was N/A, 75.9%, and 94.3% (p b 0.05), respectively for the NE, GE, and FSE groups. The kappa coefficient for GE was 0.69, indicating moderate-to-substantial agreement; the kappa coefficient for FSE was 0.93, indicating almost perfect agreement. In the NE group, 44.4% (n = 52) had an unindicated lymphadenectomy compared to 3.9% (n = 7) in the GE group and 0.3 % (n = 1) in the FSE group. 11.1% (n = 20) in the GE group and 2.7% (n = 9) in the FSE group were upstaged based on final pathology Conclusions: In this series, the specificity, sensitivity, and PPV of the FSE were very high (99%, 95%, and 99.4%, respectively). The addition of FSE to the risk-predicting diagnostic procedure decreased the rate of unindicated lymphadenectomy by 44% as compared to the NE group and by a factor 10 compared to GE only, with a decrease in the need for a second restaging surgery. FSE is an essential component of the surgical risk-predicting diagnostic procedure. doi:10.1016/j.ygyno.2011.12.383
383 Lymphovascular space invasion is an isolated poor prognostic factor for recurrence among women with intermediate to high-risk early stage endometrioid endometrial adenocarcinoma of the uterus L. Weinberg1, C. Kunos2, K. Zanotti2. 1The Cleveland Clinic Foundation, Cleveland, OH, 2Case Western Reserve - MacDonald Women's Hospital, Cleveland, OH. Objective: While previous studies have also shown lymphovascular space invasion (LVSI) to increase risk for recurrent endometrioid uterine cancer (EC), its significance relative to other known risk factors has not been defined. As such, there is some uncertainty regarding whether this finding should independently impact adjuvant treatment decisions in women with early-stage disease. Our aim was to study prognostic factors for recurrence in women with earlystage intermediate-to high-risk EC. Methods: Women with surgically treated stage I or II EC from 1994 to 2010 were included in this study if they had LVSI, FIGO grade 2 or 3 histology (G2/3), or outer half myometrial invasion (OI). Patients were excluded if their follow-up was less than 6 months. Demographic, medical, and oncologic information was obtained from chart review. We performed univariate and multivariate logistic regression analyses and Fisher's exact test for 2-way analyses of categorical variables to identify prognostic factors for vaginal, pelvic, and distant recurrences. Results: Three hundred forty-eight patients were identified, 3% were excluded. The remaining 338 patients had a median age of 68 years, a median BMI of 30.7 and a median follow-up of 60 months (range 6– 194). Seventy-three percent of patients had a lymphadenectomy. Thirty percent of patients had LVSI, 7% had cervical stromal invasion, 84% had G2/3, and 43% had OI. Adjuvant treatments included vaginal brachytherapy (61%), pelvic radiotherapy (28%), and systemic therapy (6%). The
rate of recurrence was: overall 17.2%, pelvic 10.9%, vaginal cuff 8.0%, and distant 10.7%. In univariate analyses, only LVSI and depth of myometrial invasion were significantly associated with overall recurrence. However in multivariate analysis after adjusting for age, BMI, grade, depth of invasion, cervical invasion, lymphadenectomy, and adjuvant treatment (s), only LVSI was a significant independent risk factor for total recurrence (OR = 3.07, 95% CI 1.63 -5.83). Similarly, LVSI was the only significant risk factor for all types of recurrence (pelvic, distant, and vaginal cuff; p b 0.01). Conclusions: Our study found LVSI to be an independent prognostic factor for both pelvic and distant recurrences and a better predictor than all other risk factors that define the intermediate-to high-risk category of early-stage EC. Adjuvant therapies should be considered in patients with LVSI, even in the absence of other high-risk features. doi:10.1016/j.ygyno.2011.12.384
384 The importance of applying a sentinel lymph node mapping algorithm in endometrial cancer staging: Beyond removal of blue nodes J. Barlin, F. Khoury Collado, C. Kim, D. Cassella, D. Chi, Y. Sonoda, K. Alektiar, D. DeLair, R. Barakat, N. Abu-Rustum. Memorial Sloan-Kettering Cancer Center, New York, NY. Objective: To determine the effectiveness of a surgical sentinel lymph node (SLN) mapping algorithm in detecting metastatic endometrial cancer while minimizing the need for complete lymphadenectomy (LND). Methods: A prospective database of all patients who underwent lymphatic mapping for endometrial cancer was reviewed. Cervical injection of blue dye was used in all cases. The surgical algorithm (Figure) is as follows: 1. Peritoneal and serosal evaluation and washings; 2. Retroperitoneal evaluation, including excision of all mapped SLNs and any suspicious node regardless of mapping; 3. If there is no mapping on a hemi-pelvis, a side-specific pelvic, common iliac, and interiliac LND is performed. Para-aortic LND is at attending's discretion. We retrospectively applied the algorithm to evaluate its performance. Results: Between 9/2005 and 4/2011, 498 patients received a cervical injection of blue dye for SLN mapping. At least 1 lymph node was removed in 95% of cases (474/498), and at least 1 SLN was identified in 81% of cases (401/498). There was unilateral pelvic mapping in 29% (146/498) and bilateral pelvic mapping in 51% (253/498) of cases. Over 95% (384/401) of mapped SLNs were limited to the pelvis, 4% (15/401) were in both pelvic and para-aortic regions, and there was para-aortic-only mapping in 0.5% (2/401). Median SLN count was 3 (range 1–15), and median total lymph node count was 10 (range 1–55). SLN correctly diagnosed 40 of 47 patients with nodal metastases who had at least 1 SLN mapped, resulting in a falsenegative rate of 15%. After applying the algorithm, the false negative rate for detecting nodal metastasis dropped to 2%. Only 1 patient whose lymph node spread would not have been caught by the algorithm had an isolated positive right para-aortic lymph node with a negative ipsilateral SLN and pelvic LND. Conclusions: Satisfactory SLN mapping in endometrial cancer requires adherence to a surgical SLN algorithm and goes beyond just the removal of blue SLNs. Removal of any suspicious nodes along with side-specific lymphadenectomy for failed mapping are an integral part of this algorithm. The proposed algorithm may serve as a potential middle ground in the controversy of endometrial cancer surgical staging by providing a reasonably low false, negative rate, while sparing complete bilateral LND in the majority of cases. Isolated para-aortic lymphatic spread remains a known but small risk in the majority of patients.