Adjuvant radiation for early stage endometrial cancer with lymphovascular invasion

Results Of the 126 patients, 35% ( n = 44) were treated in the early era and 65% ( n = 82) in the late era. The two groups were balanced in regards to age, race, depth of myometrial invasion, histologic grade, and cervical involvement. Significantly more patients had surgical staging and received IVRT alone in the late than early era ( p = 0.0001, 0.004, respectively). The rate of pelvic control was 93% for the early era compared to 97% for latter era ( p = 0.3). There was no significant impact of the treatment era on vaginal control, disease-free survival, or overall survival. Conclusions These data suggest that the mere presence of LVI need not trigger the use of pelvic EBRT. Instead, the decision on whether to omit EBRT in patients with LVI should be made in the context of a patient's competing risk factors and comorbid conditions. Keywords Endometrioid adenocarcinoma Early stage endometrial cancer Lymphovascular invasion Intravaginal radiation Pelvic external beam radiation Complete surgical staging Introduction Over the last decade, there has been a steady decline in the use of postoperative pelvic EBRT in patients with early stage endometrial cancer in favor of IVRT [1,2] . This was driven by the increased use of comprehensive surgical staging (CSS), and the results of two prospective randomized trials (GOG #99 and PORTEC) showing significant decrease in the rate of local relapse with adjuvant pelvic EBRT at the expense of increased risk of complications [3,4] . While the debate on whether any form of postoperative RT is needed in patients with early stage disease continues, it is clear that when adjuvant RT is recommended, most practicing physicians prefer intravaginal over pelvic RT. In the subset of patients with lymphovascular invasion (LVI), however, it remains unclear whether pelvic EBRT can be safely omitted. Several studies have shown that LVI is an independent predictor of relapse even in patients with “low-risk” stage I disease [5–13] . At the same time, the data from the Gynecologic Oncology Group (GOG) trial # 99 showed that LVI was a predictor of relapse only when combined with other poor prognostic factors such as advancing age, outer third myometrial invasion, and moderate-to-poor tumor differentiation [3] . This begs the question whether the mere presence of LVI ought to automatically prompt initiation of pelvic EBRT or, conversely, might IVRT alone suffice when LVI is evaluated in the context of other prognostic factors. To answer this question, the outcome of patients with stage IB–IIB endometrioid adenocarcinoma of the uterus with LVI was evaluated according to treatment era as the routine use of pelvic EBRT was gradually replaced by CSS and IVRT. Methods and materials An IRB-approved retrospective query was performed to capture all patients at MSKCC with a diagnosis of endometrial cancer who underwent hysterectomy between November 1988 and May 2005 and subsequently received adjuvant radiotherapy. Patient charts were reviewed and information extracted for patient age and demographics, personal medical and cancer history, surgical procedures performed, gross pathology and histology of all surgical specimens, type and dose of adjuvant radiotherapy, and disease status at last follow-up or death. Patients eligible for the analysis included those who 1) underwent a simple total hysterectomy, 2) received postoperative adjuvant radiotherapy, and 3) had a final pathologic diagnosis (confirmed in real time by a member of the MSKCC Department of Pathology) of stage IB, IC or II adenocarcinoma of endometrioid histology with documentation of the presence of LVI. Patients were excluded for any non-endometrioid histologic component at final pathologic evaluation. The above inclusion and exclusion criteria identified 126 patients eligible for analysis. The mean age was 64 years (range, 34–90 years). All patients underwent simple hysterectomy, and 85% ( n = 107) had their surgery performed by a member of the MSKCC Department of Surgery, Gynecology Service. As shown in Table 1 , the depth of myometrial invasion was ≥ 50% in 56 patients (44%), tumor histology was grade 3 in 28 (22%), and occult cervical involvement was noted in 22 (18%). Stage distribution was as follows: IB (46%, n = 58), IC (37%, n = 46), IIA (7%, n = 9), and IIB (10%, n = 13). Pelvic washings were obtained in 88% ( n = 111) of patients, pelvic lymph node sampling in 60% ( n = 76, median 13 nodes), and para-aortic lymph node sampling in 61% ( n = 77, median 5 nodes). CSS, which included pelvic and para-aortic lymph node sampling and pelvic washing, was performed in 44% ( n = 56) of the patients. None of the patients in this study had prior pelvic radiation therapy. Of the 126 patients included in the present analysis, 33% ( n = 44) received postoperative pelvic EBRT either alone ( n = 11) or with a high dose rate (HDR) IVRT boost ( n = 31). The remaining 67% ( n = 84) of patients were treated with IVRT alone. IVRT was delivered using a 2-3 cm cylinder connected to a 192 Ir HDR after-loading source. The dose was prescribed to a depth of 0.5 cm from the vaginal surface, and dose optimization was employed in all patients to ensure the lowest possible dose to the vaginal surface and introitus. When delivered alone, the median dose of HDR IVRT was 21 Gy (range: 14–21 Gy) given in 3 fractions at 2-week intervals; when combined with pelvic RT ( n = 31), the median IVRT dose was 15 Gy given in 3 fractions. Postoperative pelvic EBRT was generally delivered using 15 MV photon energy to a median dose of 50.4 Gy when used alone and 45 Gy when used in combination with IVRT. The eligible patients were divided into two groups based on the era of treatment in order to best capture the shift away from the routine use of pelvic EBRT in favor of CSS and IVRT. Patients who underwent hysterectomy between November 1988 and December 1996 were classified as early era, and those who underwent hysterectomy between January 1997 and May 2005 were classified as late era. Follow-up time was calculated from the date of hysterectomy. The follow-up schedule generally consisted of physical examinations every 3 months for the first 2 years then every 6 months for the subsequent 3 years. Pap smears were obtained every 6 months and imaging performed based on clinical findings. Associations between variables were tested using the chi-square test. Survival rates were calculated using the Kaplan–Meier method, and survival curves were compared using the log-rank test. Results Patients Of the 126 patients included in the analysis, 35% ( n = 44) underwent hysterectomy prior to 1997 (early era) and 65% ( n = 82) since 1997 (late era). The two groups were balanced in terms of age, race, tumor grade, depth of myometrial invasion, and cervical involvement ( Table 1 ). The use of pelvic EBRT was significantly higher in the early era compared to late era (55% vs. 22%; p = 0.004). Conversely, the use of CSS was significantly higher in the late era (14% vs. 61%, p = 0.0001). Furthermore, the extent of CSS varied significantly by era; only 5% of patients had > 11 pelvic lymph nodes sampled in the early era compared to 50% in the late era, p = 0.0001. The proportion of patients presenting with LVI increased between treatment eras. In the early era, patients with LVI represented 17.5% (44/252) of the total population seen at MSKCC who fulfilled the inclusion criteria for this study regardless of LVI status. This proportion increased in a statistically significant manner to 25.6% (82/320) in the late era, ( p = 0.02). The median follow-up time for the early era was 73 months compared to 40 months for the late era, but 80% (12/15) of relapses occurred within 30 months from the date of hysterectomy. Patterns of recurrence Of the 126 patients included in the present analysis, 15 (12%) experienced a relapse of disease. The rate of relapse in the early era was 18% (8/44) compared to 9% (7/82) for the late era, p = 0.2. The site of relapse included the vagina in 5 (4%) patients, pelvis in 5 (4%), para-aortic region in 9 (7%), and was distant in 10 (9%). Three of the 5 vaginal relapses were isolated events. In contrast, there were no isolated pelvic failures; all 5 pelvic relapses were associated with synchronous or brisk (within 7 months) distant relapse. Of the 10 distant relapses, 3 were to lymph nodes (axilla, supraclavicular, inguinal), 4 to lung, 2 to liver, and 1 presented as peritoneal seeding. Locoregional control The overall 5-year actuarial vaginal, pelvic and combined vaginal/pelvic control rates were 95% (95% CI: 90–99), 95% (95% CI: 91–99), and 92% (95% CI: 87–98), respectively. None of these rates varied significantly according to treatment era. The 5-year vaginal control rate was 92% (95% CI: 83–100) for the early era (1988–1996) compared to 97% (95% CI: 93–100) for the late era (1997–2005), p = 0.3. The corresponding 5-year pelvic control rates were 93% (95% CI: 84–100) and 97% (95% CI: 93–100); p = 0.3, respectively ( Fig. 1 ). Similarly, the 5-year vaginal/pelvic control at 5 years was 87% (95% CI: 76–98) for the early era and 96% (95% CI: 91–100) for the late era, p = 0.15. Since pelvic control per se is the main concern with the omission of EBRT in patients with LVI, the influence of other potential prognostic factor such as age, depth of myometrial invasion, grade, cervical involvement, and the use of pelvic RT, was analyzed. None of these factors impacted pelvic control ( Table 2 ). Even when patients were grouped as “high risk” ( n = 82) according to the GOG #99 definition [3] , there was no influence on 5-year pelvic control—95% (95% CI: 89–100) compared to 97% (95% CI: 92–100) for the low-risk group, p = 0.4. The extent of pelvic lymph node sampling was also evaluated. The 5-year pelvic control rate was 93% (95% CI: 85–100) for those without pelvic lymph nodes sampling, 93% (95% CI: 84–100) for those with ≤ 11 nodes, and 100% for those with > 11 nodes, p = 0.2. The original cohort of 126 patients was further analyzed to compare the six possible treatment sequences of the use of pelvic RT and the extent of pelvic lymph node sampling. The 5-year pelvic control rate was 100% for those who had > 11 nodes sampled irrespective of the use of pelvic RT, as well as for those with any pelvic node sampling who received pelvic RT. This is in contrast to the pelvic relapses noted when < 11 nodes were sampled and pelvic RT was omitted, though this difference did not reach statistical significance ( Table 3 ). Survival The overall 5-year actuarial rate of disease-free survival (DFS) was 85% (95% CI: 77–92). The 5-year rates of disease-specific (DSS) and overall survival (OS) were 91% (95% CI: 85–97) and 83% (95% CI: 74–92), respectively. None of these outcome parameters varied significantly according to treatment era ( Figs. 2, 3 ). The 5-year DFS rate was 80% (95% CI: 66–92) for the early era (1988–1996) compared to 89% (95% CI: 80–97) for the late era (1997–2004), p = 0.3. The corresponding 5-year DSS rates were 84% (95% CI: 73–96) and 96% (95% CI: 91–100), respectively ( p = 0.09). The influence of other prognostic factors on DFS and DSS are shown in Table 2 . Discussion The influence of LVI on outcome in patients with endometrial cancer is intriguing since it is known to increase the risk of vaginal and nodal relapse but not distant relapse [6,12,14,15] . Consequently, the presence of LVI has major implications when choosing local/regional treatment modalities such as surgery and adjuvant radiation. Despite the use of surgical staging for all patients in the GOG #99 trial, the use of adjuvant pelvic RT still conferred an added significant improvement in local/regional control for the “high intermediate risk” group of patients [3] . This leads one to question whether the presence of LVI alone should automatically trigger the need for pelvic RT irrespective of surgical staging. Clearly, it is the risk for pelvic recurrence that most concerns physicians when pelvic RT is omitted in patients with LVI even after surgical lymph node staging. Though the retrospective nature of the present study limits the ability to make broad conclusions regarding the safety of omitting pelvic RT in patients with LVI, we attempted to decrease bias in treatment choices by analyzing outcomes by treatment era. The decrease in the use of pelvic RT in favor of CSS and IVRT in the latter era did not seem to impact the outcome of patients with stage IB–IIB endometrial cancer with documented LVI. The shorter median follow-up time in the late era (40 months) compared to the early era (73 months) could obscure the detection of a difference in outcome between the two groups; however, it is important to note that 80% of the relapses in this study occurred within 30 months from of the date of hysterectomy. Though the two treatment era cohorts were balanced with regard to the recognized prognostic factors of patient age, race, tumor grade, depth of myometrial invasion, and cervical involvement, there was a statistically significant increase in the rate or reporting on the presence of LVI (17.5% for the early era vs. 25.6% for the late era; p = 0.02). This disparity raises the question whether only extensive LVI was reported in the early era. It is a limitation that the current study did not include a review of original pathology specimens (not all specimens were available for re-review) or further sub-classification of LVI according to quantity and grade [8] . The lack of a difference in pelvic control between the two treatment eras could in part be attributable to the less frequent use of surgical staging in the early cohort and any consequent undetected lymphatic micrometastases. Irrespective of treatment era, an extensive analysis of the recognized clinicopathologic prognostic factors failed to demonstrate an impact on the risk of pelvic recurrence. Because these features are not entirely independent of one another, the combined influence of several such factors as defined in the GOG #99 study [3] (increased age, outer third myometrial invasion, histologic grade 2 or 3, and LVI) was evaluated. Still, the risk of pelvic recurrence was not significantly different between those defined as high intermediate risk compared to low intermediate risk (95% vs. 97%, respectively; p = 0.4). While the shifting use of surgical staging and pelvic EBRT over time did not influence outcomes in the present study, recent data indicate that the extent of surgical lymph node staging may impact upon patients with early stage endometrial cancer [16–18] . Specifically, Cragun et al. published data on the possible therapeutic benefit of removing > 11 pelvic nodes in patients with early stage endometrial cancer and grade 3 histology [17] . Therefore, the patient cohort was divided into 6 subsets depending on whether EBRT was or was not used and on the extent of surgical staging (none vs. ≤ 11 vs. > 11 pelvic nodes sampled). Though there was no statistically significant difference in pelvic control, the patient subsets were small and events (relapse) rare. However, among the 37 LVI patients in the current study who had > 11 pelvic nodes removed and did not receive pelvic EBRT, none developed pelvic recurrence. For this subset of patients, omitting pelvic EBRT was feasible despite the fact that 21/37 had GOG #99 high intermediate risk features. The question of whether pelvic EBRT could be omitted in all surgically staged patients could only be answered through a randomized trial but that is not feasible due the prohibitively large number of patients required for such a trial. Given that ten patients in the current analysis developed extrapelvic metastatic disease, it is also important to identify those early stage patients who might benefit from adjuvant systemic therapy; this is a subject of ongoing debate. In conclusion , the data from this study suggest that the mere presence of LVI may not always portend an increased risk of pelvic relapse and, hence, need not trigger the use of pelvic EBRT, especially in surgically staged patients with > 11 pelvic nodes removed. However, the larger patient experience with pelvic RT and the lack of a head-to-head comparison to IVRT alone caution against the indiscriminate adoption of this approach in clinical practice. The decision on whether to omit pelvic EBRT in patients with LVI should be made in the context of a patient's competing risk factors and comorbid conditions. Conflict of interest statement The authors have no conflicts of interest to declare. References [1] W. Small B. Erickson F. Kwakwa American Brachytherapy Society survey regarding practice patterns of postoperative irradiation for endometrial cancer: current status of vaginal brachytherapy Int J Radiat Oncol Biol Phys 63 2005 1502 1507 [2] R.W. Naumann R.L. Coleman The use of adjuvant radiation therapy in early endometrial cancer by members of the Society of Gynecologic Oncologists in 2005 Gynecol Oncol 105 2007 7 12 [3] H. Keys J. Roberts V. Brunetto R. Zaino N. Spirtos J. Bloss 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 2004 [4] C. Creutzberg W. van Putten P. Koper M. Lybeert J. Jobsen C. Warlam-Rodenhuis The morbidity of treatment for patients with stage I endometrial cancer: results from a randomized trial Int J Radiat Oncol Biol Phys 51 2001 1246 1255 [5] J. Briet H. Hollema N. Reesink J. Aalders M. Mourits K. ten Hoor Lymphvascular space involvement: an independent prognostic factor in endometrial cancer Gynecol Oncol 96 2005 799 804 [6] A. Mariani S. Dowdy G. Keeney M. Haddock T. Lesnick K. Podratz Predictors of vaginal relapse in stage I endometrial cancer Gynecol Oncol 97 2005 820 827 [7] F. Alexander-Sefre N. Singh A. Ayhan J. Thomas I. Jacobs Clinical value of immunohistochemically detected lymphovascular invasion in endometrioid endometrial cancer Gynecol Oncol 92 2004 653 659 [8] J. Kuroki K. Hasegawa R. Kato J. Mizukawa E. Nishio H. Nishizawa Relationship between the classification of vascular invasion severity and the prognosis of uterine endometrial cancer Int J Gynecol Cancer 13 2003 47 52 [9] J. Bell A. Minnick G. Reid J. Judis M. Brownell Relationship of nonstaging pathological risk factors to lymph node metastatsis and recurrence in clinical stage I endometrial carcinoma Gynecol Oncol 66 1997 388 392 [10] R. Ambros R. Kurman Identification of patients with stage I uterine endometrioid adenocarcinoma at high risk of recurrence by DNA ploidy, myometrial invasion, and vascular invasion Gynecol Oncol 45 1992 235 239 [11] C. Feltmate L. Duska Y. Chang C. Flynn N. Nikrui E. Kiggundu Predictors of recurrence in surgical stage II endometrial adenocarcinoma Gynecol Oncol 73 1999 407 411 [12] A. Mariani M. Webb G. Keeney G. Aletti K. Podratz Predictors of lymphatic failure in endometrial cancer Gynecol Oncol 84 2002 437 442 [13] G. Pitson T. Colgan W. Levin G. Lockwood L. Manchul M. Milosevic Stage II endometrial carcinoma: prognostic factors and risk classification in 170 patients Int J Radiat Oncol Biol Phys 53 2002 862 867 [14] A. Mariani M.J. Webb G.L. Keeney T.G. Lesnick K.C. Podratz Surgical stage I endometrial cancer: predictors of distant failure and death Gynecol Oncol 87 2002 274 280 [15] D.E. Cohn N.S. Horowitz D.G. Mutch S.M. Kim T. Manolitsas J.M. Fowler Should the presence of lymphvascular space involvement be used to assign patients to adjuvant therapy following hysterectomy for unstaged endometrial cancer? Gynecol Oncol 87 2002 243 246 [16] C.V. Lutman L.J. Havrilesky J.M. Cragun A.A. Secord B. Calingaert A. Berchuck Pelvic lymph node count is an important prognostic variable for FIGO stage I and II endometrial carcinoma with high-risk histology Gynecol Oncol 102 2006 92 97 [17] J.M. Cragun L.J. Havrilesky B. Calingaert I. Synan A.A. Secord J.T. Soper Retrospective analysis of selective lymphadenectomy in apparent early-stage endometrial cancer J Clin Oncol 23 2005 3668 3675 [18] J.K. Chan R. Urban M.K. Cheung J.Y. Shin A. Husain N.N. Teng Lymphadenectomy in endometrioid uterine cancer staging: how many lymph nodes are enough? A study of 11,443 patients Cancer 109 2007 2454 2460