Introduction

Pelvic exenteration (PE) has remained a controversial topic since its original description by Brunschwig in 1948.1 The original procedure described en bloc resection of the internal and external reproductive organs, the bladder with the urethra for females and the bladder with the prostate for males, the pelvic ureter, the rectum, and the sigmoid colon. Although Brunschwig himself described the procedure as “brutal and cruel,” he believed that it did “save lives.”1,2 Prior to the development of the ileal conduit by Bricker for urinary diversion, perioperative morbidity rates approached 90 %.1,3 Moreover, early perioperative mortality rates exceeded 23 %, and no patient survived the past 5 years. By the late 1960s, the medical community believed PE was of little benefit and only served to cause cancer patients disfigurement and further suffering.2 Despite these concerns, PE has evolved and continues to play an important role in the management of advanced pelvic malignancies, though its effectiveness has never been fully validated with controlled clinical trials.

General indications for pelvic exenteration include locally advanced or recurrent bladder and rectal cancers and salvage therapy for central recurrence of gynecologic cancers following primary radiotherapy. Although PE has slowly gained acceptance over the past 50 years, it remains a subject of heated debate in many surgical specialties focused on whether the oncologic effectiveness outweighs the reported morbidity and mortality of the procedure. Previous studies regarding PE for the treatment of pelvic malignancies have been limited by small, single-institution case series that were collected during lengthy periods of time, thereby encompassing different eras of surgical evolution.4–11 As such, to date, there are no studies of considerable size assessing outcomes of patients after pelvic exenteration for advanced pelvic malignancies.

The purpose of this study was to independently query two large national databases to (1) estimate the 30-day complication and mortality, (2) find predictors of complications, and (3) evaluate long-term survival after PE for patients with advanced colorectal and bladder cancers. Cervical cancers were not included, as PE for cervical cancer is most commonly utilized as salvage therapy for recurrence following primary treatment with XRT. The use of large national clinical databases permits larger sample sizes for a contemporary analysis than is possible from traditional single-center data sources and allows for more up-to-date and generalizable conclusions to be drawn.

Methods

Data from the 2005 through 2011 American College of Surgeons (ACS) National Surgical Quality Program (NSQIP) Participant User File were used to estimate contemporary 30-day morbidity and mortality rates after pelvic exenteration. The NSQIP PUF contains information on 135 perioperative variables for patients undergoing surgical procedures at one of over 500 participating academic and community hospitals.12,13 Variables captured in this database included preoperative risk factors for intra- and postoperative complications, intraoperative variables, and 30-day postoperative mortality and morbidity. This information is prospectively collected at each participating institution by an ACS-trained on-site surgical clinical reviewer and is subject to a formal auditing process to ensure data quality.

In this study, we defined pelvic exenteration as a major open surgery that included en bloc removal of the bladder, rectum, and prostate in males and bladder, rectum, and female reproductive organs in females. To achieve this, all patients with principal Current Procedural Terminology (CPT) codes for pelvic exenteration (anterior, posterior, and complete) were included in this study (45126 or 51597). The primary outcome variables were 30-day mortality and 30-day composite variables for any complication (yes/no) and major complication (yes/no). We defined any complication to include all postoperative complication variables available in the NSQIP PUF, whereas a major complication was defined as all complication variables with the exception of superficial surgical site infection, urinary tract infections (UTI), deep venous thrombosis, peripheral nerve injury, or graft failure.

To address our third objective, which was to estimate overall survival after pelvic exenteration, the Surveillance, Epidemiology, and End Results (SEER) database was queried for all patients diagnosed between 2004 and 2010 with stage T4, M0, N0–N2 pelvic malignancies based on AJCC TNM criteria (6th edition 2004) who were treated with pelvic exenteration surgery using seer specific coding. SEER is sponsored by the National Cancer Institute and has been used to track cancer incidence and patient survival since 1973. The SEER database currently covers approximately 28 % of the US population and captures 98 % of all cancer cases within the surveyed geographic areas.

Statistical Analyses

The distributions of preoperative, intraoperative, and postoperative variables were described with medians and proportions. To estimate the predictors of major morbidity after pelvic exenteration, a logistic regression model was fitted with stepwise forward regression using a two-sided significance level to stay in the model of 0.20. The candidate predictors were BMI, gender, age, American Society of Anesthesiologists (ASA) classification, preoperative creatinine, preoperative hematocrit, diabetes mellitus, tobacco use (defined as documented smoking within the year prior to surgery), preoperative dyspnea, coronary artery disease, hypertension requiring medication, COPD, pre-existing wound infection, chronic steroid use, preoperative transfusion requirement, preoperative SIRS/sepsis/septic shock, bleeding disorder, chemotherapy within 30 days prior to operation, and radiation therapy within 90 days prior to surgery.

Overall survival (OS) was defined as the time from surgery to death due to any cause. Patients were stratified by type of malignancy (rectal and bladder), and the Kaplan–Meier method was used to describe survival, both overall and stratified by disease stage.14 A nonoperative control group was defined using the same SEER histology and staging information utilized to identify our primary PE cohort. For the control group, we considered only patients with a reason for no cancer-directed surgery of “Patient or patient's guardian refused” as recorded in SEER. Based on SEER definitions, patients were at least deemed potential surgical candidates by their providers, and the intent of this strategy was to avoid biasing our results toward a benefit to PE as a result of patient selection. We made an affirmative decision to control for type I error at the level of the comparison. A p value ≤0.05 was used to indicate statistical significance for all comparisons and analyses. All statistical analyses were done using R version 3.0.1 (Vienna, Austria).

Results

The ACS-NSQIP database contained 377 patients who had undergone pelvic exenteration between 2005 and 2011. These patients were classified according to CPT codes into 226 rectal (59.9 %) and 151 urological (40.1 %) indications. The distributions of preoperative and intraoperative variables are given in Table 1 stratified by primary malignancy type. The median age of patients undergoing pelvic exenteration was 62 (IQR 52 to 71), with patients in the urological group being slightly older. Overall, a larger proportion of patients in the colorectal group had received radiotherapy within 90 days prior to PE surgery (31.9 versus 5.4 %, p < 0.001), while preoperative chemotherapy rates were similar (12.6 versus 10.8 %). The median operative time for PE approached 7 h (414 min) and was 104 min longer in the colorectal group (470 versus 366 min).

Table 1 Distribution of preoperative and intraoperative patient characteristics from the NSQIP patient sample
Full size table

Postoperative outcomes of patients after PE are reported in Table 2. Incidence of postoperative mortality was low across all operative groups (1.3 %). The majority of patients (60.7 %) suffered at least one postoperative complication. Major complication and reoperation rates were also high (49.9 and 9.5 %, respectively) but similar across operative indications. Overall, the patients undergoing PE for colorectal indications seemed to have higher rates of infectious complications [deep surgical site infection (SSI), organ space SSI, and sepsis] but lower incidences of postoperative bleeding requiring transfusion. The median length of hospitalization was 9.5 days (IQR 7 to 15) and was slightly longer in the colorectal group (10 versus 9 days, p = 0.01). The most common complications after PE were superficial SSI (12.7 %), postoperative bleeding requiring transfusion (32.1 %), sepsis (10.6 %), and UTI (10.6 %).

Table 2 Unadjusted postoperative outcomes after pelvic exenteration based on the NSQIP sample
Full size table

Given the high incidence of major complications after PE, we wished to determine which preoperative variables were associated with major postoperative complications. The results of the forward stepwise logistic regression model are shown in Fig. 1. Specifically, patients with ASA classification of three or greater [adjusted odds ratio (AOR), 1.66 (1.0–2.76), p = 0.05] and those requiring preoperative transfusion [AOR, 3.38 (0.57–20.1), p = 0.18] appeared to be at increased risk of major postoperative complications. Patients with normal preoperative creatinine levels [AOR, 0.22 (0.12–0.41), p < 0.001] those treated with preoperative radiation within 90 days of surgery [AOR, 0.59 (0.33–1.05), p = 0.07] and those with a history of at least two alcoholic drinks per day [AOR, 0.20 (0.04–1.0), p = 0.051] had a reduced risk of major postoperative complications.

Fig. 1
figure 1

Predictors of major complication after pelvic exenteration after multivariable adjustment

Full size image

In an attempt to estimate the oncologic effectiveness of PE for locally advanced pelvic malignancy, OS stratified by PE versus no PE for patients with T4M0 rectal and bladder malignancies were estimated using the SEER database and are shown in Figs. 2 and 3, respectively. For the 691 patients with T4M0 rectal cancer, median survival for the PE group was 48 months (95 % CI, 38–58 months) compared to 10 months (95 % CI, 4–13 months) in the nonoperative control group that refused surgery (p < 0.001). Patients undergoing pelvic exenteration for T4M0 bladder cancer (n = 389) had significantly worse overall median survival of 14 months (95 % CI, 12–16 months) following PE, though this was substantially higher than the median survival of 1.5 months among the nonoperative control group (p < 0.001). On subgroup analysis shown in Fig. 4 a, b, the nonoperative patients similarly had markedly worse survival, regardless of procedure type and stage of disease (p < 0.001 in both instances).

Fig. 2
figure 2

Kaplan–Meier survival for locally advanced T4 colorectal cancer, pelvic exenteration versus nonoperative management

Full size image
Fig. 3
figure 3

Kaplan–Meier survival for locally advanced T4 bladder cancer, pelvic exenteration versus nonoperative management

Full size image
Fig. 4
figure 4

Kaplan–Meier survival for locally advanced a colorectal cancer and b bladder cancer stratified by disease stage and use of pelvic exenteration

Full size image

Discussion

Despite modern advances in surgical techniques, anesthesia, and postoperative care,5,6,10,11 our findings suggest that PE continues to be associated with high postoperative morbidity with half of all patients who undergo PE experiencing a major post-surgical complication. However, we also demonstrated that 30-day mortality secondary to PE was less than 2 %. Assuming PE provides an oncologic or palliative benefit, these results would support PE as an acceptable treatment option for patients willing and deemed medically fit to endure a likely postoperative complication.

While we report considerable morbidity and mortality following PE, these results do call into question recent single-institution studies describing prohibitively high postoperative mortality and morbidity rates (17 and 100 %, respectively).7,9,10 Furthermore, 5-year survival rates reported in the literature have been highly variable (20–55 %) and difficult to interpret as they have been derived from small, uncontrolled case series.6–11,15 To the best of our knowledge, this is the largest and most contemporary review of surgical and oncological outcomes following PE for bladder and rectal cancers. We report that PE is an effective technique with a high percentage of long-term survivors and suggest that preconceived notions of surgical contraindications such as advanced age, lymph node metastasis, or palliative intent may not be valid. Prior to this study, the largest report included 282 patients from the Hospital of Zurich, with substantial morbidity.16 However, a group at the Ellis Fischel Cancer Center described encouraging results similar to ours, reporting 50-year follow-up for 238 patients, although their study was limited in scope to one center and potential selection bias.17 Our overall survival estimates suggest that PE continues to have a role in the management of locally advanced pelvic malignancies.

Although PE has been the primary procedure indicated for salvage therapy following central recurrence of cervical cancer after primary or adjuvant radiation and chemotherapy,18,19 its role in primary, locally advanced cervical cancer remains unclear. Because PE in cervical cancer is usually only indicated as salvage therapy after recurrence following XRT, we chose to only include colorectal and urologic malignancies in this analysis investigating the role of PE as a treatment option for locally advanced pelvic malignancy.

Our finding that patients treated with preoperative radiation within 90 days of surgery had lower risk of complications is somewhat counter intuitive. We reason that this can be explained by additional measures taken during surgical cases treated with adjuvant radiation therapy to prevent wound complications such as myocutaneous flap or pelvic floor reconstruction that are not accounted for in the NSQIP database. Such precautions are taken with regularity at our institution for surgical candidates who have received preoperative radiation therapy. In addition, numerous studies have found that neoadjuvant radiation was associated with a marked reduction in tumor size and decreased incidence of positive margins, which improved perioperative morbidity.20–22 Regarding preoperative alcohol use having a protective benefit, we assume this is due to a strong selection bias, where of those patients with a significant drinking history, only those select individuals who appeared particularly healthy were offered PE.

Limitations of this study included lack of data regarding patterns of recurrence after pelvic exenteration, types and duration of adjuvant chemo- and radiotherapy, and the retrospective nature of both SEER and NSQIP data. It is possible that significant selection bias played a substantial role with respect to both the NSQIP and SEER databases in that only those patients deemed the most suitable surgical candidates were offered PE. Furthermore, given our analysis strategy of querying CPT codes, we were likely unable to identify all PE cases in that some combination surgeries such as colectomy plus cystectomy may have been missed, and similarly, we were unable to stratify our analyses by anterior, posterior, and total exenterations. Lastly, a systematic lack of post-procedural quality of life (QOL) data limits our ability to adjust for patient-centered factors that might affect the decision of whether or not to pursue PE as a treatment option. Recent quality of life studies suggest that PE patients do indeed have good QOL compared to those patients who underwent routine resection of a rectal primary and also the general population.23 For patients with late stage rectal or bladder cancer, PE may improve survival and offer a sense of continued hope. Moreover, modern PE, in selected cases, may have value as a palliative procedure, having been reported to effectively treat disabling sequelae of pelvic cancers such as pain, tumor abscesses, recurrent hemorrhage, bowel obstruction, and enterourinary or genitourinary fistulas.19,24

Conclusion

Our findings, using contemporary national datasets, provide the highest level of evidence to date in support of PE for the management of advanced colorectal and urologic malignancies. Future large prospective studies are necessary to evaluate the differential role of PE in primary versus recurrent disease and to validate the results presented here. However, until such investigations become feasible, we believe the results of this study provide strong evidence to support pelvic exenteration as a viable treatment option for advanced pelvic malignancies in appropriately selected patients.