Predictive Factor of Conversion to Laparotomy in Minimally Invasive Surgical Staging for Endometrial Cancer.

There were 788 cases of endometrial cancer that underwent hysterectomy-based surgical staging during the study period. Of those, 29 cases (3.7%) had surgery at an outside institution, and the remaining 759 cases had surgical staging within our institution. Of these, there were 251 women(33.1%) with endometrial cancer who underwent an attempt at TLH. There was a trend toward an increasing rate of TLH attempt during the study period (rate of TLH attempt per annual case, 0% for 2000–2005, 3.9%–14% for 2006–2008, 35.1%–64.4% for 2009–2011, and 48.8%–69.2% for 2012–2014, P < 0.001, Supplemental Digital Content 1 http://links.lww.com/IGC/A325 ). These 251 cases of endometrial cancer attempted for TLH- based surgical staging comprised the study population. There were 30 cases (12.0%; 95% CI, 7.9–16.0) that resulted in laparotomy conversion among 251 cases. The most common reason for laparotomy conversion was a large uterus limiting visualization (26.3%), followed by intrapelvic adhesions limiting surgical ability (23.7%), poor visualization due to adiposity (13.2%), patient respiratory intolerance to Trendeleburg positioning (10.5%), intraoperative vascular injury (10.5%), intraoperative urinary tract injury (10.5%), and laparoscopic findings of intra-abdominal metastatic disease (5.3%). There were 8 cases (26.7%) with multiple reasons for laparotomy conversion. Most laparotomy-conversion cases had a vertical midline incision as the type of skin incision (80%), followed by Pfannenstiel incision (16%) and extended port site incision (4%). Patient demographics are shown in Table 1 . In the entire cohort, mean age was 53.8 years, and most study population was Hispanic (71.3%) and obese (BMI ≥30 kg/m 2 ; 64.9%). Approximately one fourth (26.3%) of cases were morbidly obese. More than 95% of the study population had mild to severe systemic disease (ASA performance score, ≥2; 95.6%), and one fourth (25.5%) of cases had severe systemic disease (ASA performance score ≥3; 25.9%). Among the tested variables for patient demographics, laparotomy conversion was significantly associated with larger BMI (mean BMI for conversion vs nonconversion, 40.6 vs 34.3 kg/m 2 ; P = 0.025). Physical examination provided useful information for laparotomy conversion, and an enlarged uterus (20.8%) and suboptimal physical examination due to patient large body habitus (19.1%) were associated with laparotomy conversion when compared with small/normal size uterus (4.5%, P < 0.001). History of major intra-abdominal surgery was not associated with laparotomy conversion ( P = 0.84). Factors related to tumor were examined ( Table 2 ). In the entire cohort, the majority were endometrioid type (87.6%), grade 1 tumor (66.5%), and stage 1 disease (86.1%). Median uterine size was 161 g. When compared with the nonconversion group, the laparotomy-conversion group was associated with a higher prevalence of elevated CA-125 (≥35 vs <35 IU/L, 23.7% vs 10.0%, P = 0.028), higher grade (1, 2, and 3, 10.8%, 5.0%, and 22.7%, respectively, P = 0.032), higher stage ( I , II, III, and IV, 8.8%, 14.3%, 29.2%, and 75.0%, respectively, P < 0.001), deeper myometrial invasion (none, inner third, mid third, and outer third, 9.7%, 10.9%, 7.7%, and 29.6%, respectively, P = 0.025), presence of lymphovascular space invasion (no vs yes, 10.1% vs 24.2%, P = 0.038), and larger uterine size (<250 vs ≥250 g, 10.0% vs 28.1%, P = 0.014). Factors relating to the surgeon to predict laparotomy conversion were examined ( Table 3 ). In the entire case, most cases were done in the recent study period (2011–2014, 74.9%) and started in the morning (70.1%). There were in total of 9 attending surgeons who performed TLH during the study period. Most cases were assisted by graduating year fellows-in-training (51.8%) or graduating resident physicians (76.1%). Pelvic and aortic lymphadenectomy were performed in 36.7% and 11.2% of cases, respectively. Laparotomy conversion was significantly associated with pelvic lymph- adenectomy (no vs yes, 6.9% vs 20.7%, P = 0.002), aortic lymphadenectomy (8.5% vs 39.3%, P = 0.001), and omentectomy (10.1% vs 46.2%, P = 0.002). Chronologic and staff experience factors were not associated with laparotomy conversion in our study. Outcomes of laparotomy conversion were examined ( Table 4 ). In the entire cohort, median EBL, operating time, and length of hospital stay were 100 mL, 237 minutes, and 2 days, respectively. There was an inverse correlation between chronologic year of surgery and operating time (Spearman r = −0.20; P = 0.002). Rates of readmission and intraoperative major complication in our study population were 2.4% and 4.0%, respectively. Cases resulting in laparotomy conversion were significantly associated with larger EBL (conversion vs nonconversion, 350 vs 100 mL, P =0.001), increased incidence of intraoperative transfusion (16.7% vs 2.3%, P = 0.003), longer operating time (333 vs 224 minutes, P < 0.001), longer hospital stay after surgery (4 vs 2 days, P = 0.001), higher readmission rate within 30 days after surgery (10.0% vs 1.4%, P = 0.024), and higher intraoperative major complication rates (16.7% vs 2.3%, P = 0.003). Among the specific types of major complications, laparotomy conversion was associated with vascular injury (laparotomy conversion vs nonconversion, 10.0% vs 0%, P = 0.002), intestinal injury (6.7% vs 0.5%, P = 0.038), and urinary tract injury (13.3% vs 0.9%, P = 0.002). All 4 urinary tract injuries in the laparotomy-conversion group were at the time of hysterectomy, and in 3 (75%) of 4 cases, the urinary tract injury occurred before the conversion. Among 11 patients who underwent para-aortic lymphadenectomy in the laparotomy-conversion group, 2 cases (18.2%) underwent para-aortic lymphadenectomy after laparotomy conversion. The remaining 9 patients (81.8%) underwent para-aortic lymphadenectomy via a laparoscopic approach, and there were 2 patients (22.2%) who had a vascular injury during the laparoscopic approach that required laparotomy conversion to repair the injury (1 aortic injury and 1 vena cava injury). Among 19 patients who underwent pelvic lymphadenectomy in the laparotomy- conversion group, 10 patients (52.6%) underwent pelvic lymphadenectomy before laparotomy conversion, and the remaining 9 (47.4%) patients underwent pelvic lymphadenectomy via laparotomy approach. Laparotomy conversion had a significantly higher prevalence of intensive care unit admission after surgery (6.7% vs 0%, P = 0.014). Multivariate analysis was performed to determine the independent risk factors for laparotomy conversion. Eleven statistically significant covariates from patient, tumor, and surgeon factors in univariate analysis were initially entered in the model. With a conditional backward method, there were 5 variables that remained as the independent risk factors associated with laparotomy conversion in the final model ( Table 5 ). These independent risk factors included enlarged uterus or a uterus that was unable to be assessed because of patient large body habitus during preoperative examination (OR, 3.55; P = 0.034), BMI of 40 kg/m 2 or greater (OR, 4.51; P = 0.011), stages III and IV disease (OR, 4.69; P = 0.012), uterine size 250 g or greater (OR, 3.49; P = 0.026), and para-aortic lymphadenectomy (OR, 10.5; P = 0.001). The following numbers of risk factors were positively correlated with risk of laparotomy conversion: none, 1.1%; single risk factor, 2.6% (OR, 5.0; 95% CI, 0.55–45.7; P = 0.15); double risk factors, 21.7% (OR, 24.9; 95% CI, 3.16–196; P = 0.002), and triple or more risk factors, 50% (OR, 90.0; 95% CI, 10.7–755; P = 0.001; Fig. 1A ). Specific combination patterns of these risk factors and laparotomy-conversion rates are shown in Supplemental Digital Content 2 , http://links.lww.com/IGC/A325 . The number of predictive factors for laparotomy conversion was correlated to the outcome and complications of TLH ( Supplemental Digital Content 3 , http://links.lww.com/IGC/A325 ). Multiple risk factors were significantly associated with large EBL, long operating time, and long hospital stay (all, P < 0.001). Multiple predictive factors were also associated with vascular and urinary tract injury (both, P < 0.01). Because a large uterus was the most common reason for laparotomy conversion, actual uterine weight was estimated by using the results of transvaginal ultrasonography performed before the surgical staging. There were 176 cases available for the results. In a linear regression model, ultrasonographic 3-dimensional volume of 496 cm 3 in preoperative uterine size correlated with actual uterine weight of 250 g: uterine weight (g) = 61.5 + 0.38 × 3-dimensional uterine volume in preoperative ultrasonogram (cm 3 ), P < 0.001, Figure 1B .

After institutional review board approval was obtained at University of Southern California, a retrospective collection of information from an institutional database of endometrial cancer cases was used for evaluation. First, all the cases that underwent hysterectomy-based surgical staging at Los Angeles County Medical Center between January 1, 2000 and May 30, 2014 were identified. Then, medical records of the identified cases were examined. Cases completed with minimally invasive approach (nonconversion group) and cases initiated with minimally invasive approach but ended in abdominal hysterectomy via laparotomy conversion (laparotomy-conversion group) were eligible for statistical analysis. Minimally invasive approach for surgical staging in our study refers to total laparoscopic hysterectomy (TLH). At our institution, a robotic system was not available at the point of data analysis. Laparotomy conversion was defined as a hysterectomy that initially started with laparoscopy but required subsequent laparotomy to complete surgical staging or to repair surgical complications associated with surgical staging. Subsets of cases in this study were within the context of our previous studies. 12 – 14 Among cases eligible for statistical analysis, medical records were further examined to extract the following variables of interests: (1) patient factors, (2) tumor factors, and (3) surgeon factors. (1) For patient factors, the following variables were obtained from preoperative workup records at the time of cancer diagnosis: patient age, ethnicity, pregnancy history including number of vaginal deliveries, BMI, medical and surgical history, findings from physical examination, 3-dimensional uterine size in transvaginal ultrasonography, preoperative laboratory values, and performance status (American Society of Anesthesiology [ASA] physical status). 15 (2) For tumor factors, the following information was obtained from the pathology reports of the surgical staging procedure: histologic subtype, tumor grade, cancer stage, uterine weight, depth of myometrial invasion, presence of lymphovascular space invasion, and number of lymph nodes sampled. Based on the American Board of Obstetrics and Gynecology classification, a uterine weight of 250 g was chosen as a cutoff for high uterine weight. (3) Surgeon factors represent the following setting and details of surgery for hysterectomy-based surgical staging: chronological time and date of surgery (year, month, and time), surgeon’s year of training after graduation (attending, fellow, and resident), reason for laparotomy conversion, intraoperative findings, surgical complication, estimated blood loss (EBL), operating time (skin incision to closure), disposition after the surgery, length of hospital stay, and presence of readmission within 30 days from surgical staging. Our primary interest of analysis was to identify the risk factors of laparotomy conversion among the cases that underwent TLH for endometrial cancer by comparing variables between the following 2 groups: the nonconversion group and the laparotomy-conversion group. Our secondary interest of analysis was to construct a predictive model of laparotomy conversion based on the extent of independent risk factors. In addition, systematic literature search was performed to examine the rationale of laparotomy-conversion rates in surgical staging for endometrial cancer. Continuous variables were assessed for normality (Kolmogorov-Smirnov test) and expressed as appropriate (mean with standard deviation or median with range). Student t test or Mann-Whitney U test was performed for continuous variables as appropriate. Categorical variables were evaluated with Fisher exact test or χ 2 test as appropriate, expressed with odds ratio (OR) and 95% confidence interval (CI). Spearman correlation coefficient was used for continuous and ordinal variables. Among statistically significant variables between the nonconversion group and the laparotomy-conversion group by the previously mentioned univariate analysis, multivariable analysis with a binary logistic regression model (conditional backward method) was further performed to determine independent risk factors of laparotomy conversion. A linear regression model was used to estimate the actual uterine weight by using the preoperative transvaginal ultrasonogram. A P value of less than 0.05 was considered statistically significant (all 2-tailed). Statistical Package for Social Sciences (SPSS, Version 12.0, IL) was used for all analyses.

The outcome of minimally invasive surgery for endometrial cancer consists of a complex triad between patient, tumor, and surgeon factors. Our study systematically examined these factors and found that there are certain types of independent risk factors associated with increased risk of laparotomy conversion. Because laparotomy conversion was associated with compromised surgical outcomes of patients with endometrial cancer when compared with those with a completed laparoscopic approach, proper selection and decision of patients with endometrial cancer for a laparoscopic approach by adopting our predictive model of laparotomy conversion will be useful when surgical staging is conducted. To provide the rational of laparotomy conversion in endometrial cancer surgical staging, a systematic literature review was performed by searching public search engines PubMed and MEDLINE on November 30, 2014. The entry keywords were “endometrial cancer,” “laparoscopy,” and “conversion.” Eligibility criteria were that the original study examined conventional TLH for endometrial cancer (not robotic-assisted) and reported the laparotomy-conversion rate. Year of publication, country of study, subject size for laparoscopic hysterectomy group, BMI, laparotomy-conversion rate, reported EBL and transfusion rate, length of hospital stay, and intraoperative complication were extracted. There were 31 studies including our current study examined for the analysis. 5 – 10 , 16 – 39 The total number of cases examined for the review analysis was 5899. The most common geographic area of study was North America (45.2%), followed by Europe (35.5%). Most studies were reported in 2005 or later (80.6%). Median (range) of age and sample size reported in the studies were 61 (54–76) and 103 (20–1682) cases, respectively. Median (range) of reported BMI and laparotomy-conversion rate were 29 (24–40) kg/m 2 and 3.8% (0%−36.4%), respectively. There was a trend of decreasing laparotomy-conversion rate in recent years (r = −0.44; P = 0.014; Fig. 1C ). There was a significant positive correlation between laparotomy-conversion rate and BMI (r = 0.46; P =0.023; Fig. 1D ) as well as laparotomy-conversion rate and age (r = 0.39; P = 0.037; Fig. 1E ). Study size was not correlated with laparotomy-conversion rate (r = 0.30; P = 0.11). When our study is compared with other studies, our results did not show a downtrend of laparotomy-conversion rate during the study period, whereas the systematic review pointed toward a downtrend of laparotomy-conversion rate ( Fig. 1C ). During our study time, there was an increase in TLH attempts for endometrial cancer cases in our study ( Supplemental Digital Content 1 http://links.lww.com/IGC/A325 ). Although large body habitus in our study population (mean BMI, 35.1 kg/m 2 ) compared with other study populations could contribute to the higher laparotomy-conversion rate, there is a possibility that there may be a tendency or a low threshold or nonselective choice of scheduling laparoscopic approach in endometrial cancer surgical staging in recent years in our study population. Although recent literature points toward a beneficial outcome for minimally invasive approach more than laparotomy in endometrial cancer management, there is a scarcity of evidence that exists in relation to predictive factors when it comes to the laparotomy conversion in such an approach. 5 , 6 , 9 , 40 Large BMI and old age seem to be related to laparotomy conversion in our systematic review ( Figs. 1D – E ); however, few other predictive factors have been reported in the past literature. 5 , 10 Therefore, risk factor–based theoretical selection as shown in our results rather than pan selection of minimally invasive approach for endometrial cancer surgical staging would be a reasonable strategy to reduce and avoid the complications related to laparotomy conversion and to maximize the surgical outcome of patients with endometrial cancer. Similarly, if a patient with multiple risk factors preoperatively underwent a minimally invasive approach and the surgeon encountered a technical challenge during surgery, earlier rather than later decision making for laparotomy conversion is a practical tactic. Large uterine size was the most common indication for laparotomy conversion that was associated as an independent predictor of laparotomy conversion in our study. Specifically, uterine weight of 250 g or greater significantly increased the risk of laparotomy conversion 3-fold higher than those with uterine weight less than 250 g (28% vs 10%, Table 5 ). Thus, predicting actual uterine weight before surgical staging will help guide clinicians and provide more ideas for surgical planning. Our study found that 3-dimensional transvaginal ultrasonography is a useful tool to estimate the actual uterine weight ( Fig. 1B ). Patients with endometrial cancer are often morbidly obese and the bimanual vaginal examination can be challenging to estimate the uterine size. Further studies are warranted to validate the value of transvaginal ultrasonography in the assessment of surgical staging in endometrial cancer. A strength of this study was that this is one of largest studies examining the conversion rate in laparoscopic surgical staging in endometrial cancer (top 20 percentile among reported in 31 studies examined for systematic review). This study examined a total of 34 variables in a comprehensive manner with decent data quality (very little missing data). A weakness of our study is that this is a retrospective study and there may be possible confounding factors missed in our data. For instance, the exact discussion and indication for allocating to minimally invasive approach versus laparotomy at the time of surgical planning were not retrievable from the medical records. In summary, a minimally invasive approach is becoming a common treatment strategy for surgical staging in patients with endometrial cancer. Because laparotomy conversion in surgical staging can result in compromised surgical outcomes, awareness of risk factors that contribute to laparotomy conversion is a crucial consideration for surgical management in patients with endometrial cancer. Prospective studies based on risk stratification of laparotomy conversion merits further investigation.