Impact of bowel obstruction at the time of initial presentation in women with ovarian cancer

Authors


Dr AB Olawaiye, Department of Obstetrics, Gynecology and Reproductive Sciences, Division of Gynecologic Oncology, Magee-Women’s Hospital, University of Pittsburgh Medical Center, 300 Halkett Street, Pittsburgh, PA 15213, USA. Email olawaiyea@mail.magee.edu

Abstract

Objective  To determine whether the presence of bowel obstruction at the time of initial presentation has any prognostic significance in these women.

Design  Retrospective cohort study.

Setting  Dedicated gynaecological oncology service of a large tertiary institution.

Population  Women who had a bowel obstruction as part of their initial presentation of ovarian cancer were identified between 1995 and 2007. Each woman was matched with four control women (with disease but no obstruction).

Methods  Women with disease were compared with controls to determine the impact, if any, of bowel obstruction at presentation. Several prognostic variables including bowel obstruction were also evaluated in a Cox proportional hazard model.

Main outcome measures  Progression-free survival (PFS) and overall survival (OS).

Results  Forty-eight women with disease and 192 controls were identified during the study period. The median follow-up period was 19 months among women with disease versus 20 months in controls. No differences were seen in demographics and clinical characteristics of the women. Optimal cytoreduction rate was similar between the two groups (75% versus 78%, = 0.7). Patients with bowel obstruction had a shorter PFS and OS compared with controls [19 months versus 21 months (= 0.01) and 22 versus 35 months (P = 0.008)], respectively. Bowel obstruction at presentation was an independent prognostic variable with a hazard ratio of 1.5 (= 0.009). Other prognostic variables were age, stage and extent of surgical cytoreduction.

Conclusions  Bowel obstruction at the time of initial presentation is an adverse prognostic factor in women with ovarian cancer.

Introduction

Ovarian cancer remains the most lethal of gynaecological malignancies in developed countries,1 with a US estimate of 21 650 diagnoses and 15 520 deaths in 2008.2 The poor survival associated with this disease is largely related to most women presenting with advanced-stage disease. Aggressive surgical cytoreduction and platinum/taxane combination chemotherapy have resulted in improved outcomes for many women with ovarian cancer since 2000. The rationale for aggressive tumour resection is based on a number of retrospective studies that indicate that complete resection of all visible disease is associated with a more favourable survival outcome.3–8 In fact, maximum initial cytoreductive surgery was found to be the most powerful determinant of survival in a recent meta-analysis.9

Ovarian cancer commonly spreads along the peritoneal surfaces in the abdomen and commonly involves the bowel.10 Accordingly, symptomatic ovarian cancer often suggests the presence of direct or indirect bowel involvement. Bowel obstruction, however, is an uncommon initial presentation. A small number of studies have estimated that 8–30% of women with ovarian cancer present with bowel obstruction.11,12 Although the aetiology is multifactorial, there are distinct mechanisms that provoke malignant bowel obstruction including extension of the tumour directly into the bowel wall; adjacent, impinging carcinomatosis; and adherence of tumour to the external bowel surface.13

The identification of prognostic factors predictive of poor outcome in women with ovarian cancer should have important therapeutic implications. In women with advanced epithelial ovarian cancer, several studies have identified age, performance status, histological cell type, stage, histological grade, residual tumour size and presence of ascites as independent prognostic factors.14–17 None of the previous reports have investigated bowel obstruction at presentation as a prognostic factor. Therefore, this investigation presents the experience in a single institution with women who have ovarian cancer and present with bowel obstruction in an attempt to characterise the impact on prognosis, if any, of bowel obstruction at the time of primary presentation. In addition, we hypothesised that bowel obstruction at first presentation in women with ovarian cancer does not have a negative impact on the ability to achieve optimal cytoreduction.

Methods

An institutional review board-approved retrospective analysis of the Cancer Registry database at our institution was performed. All women who had their primary surgery in our institution and pathologically confirmed FIGO Stage III or IV epithelial ovarian, fallopian tube and peritoneal cancer between January 1995 and December 2007 were identified. All operations were performed by gynecologic oncology faculty, with intent to achieve optimal cytoreduction. Optimal cytoreduction was considered achieved when there was no residual tumour nodule more than 1 cm in its largest diameter at the end of primary operation. Chemotherapy was largely platinum-based and reflected standard protocols used during the study period. Patients were excluded from the study for the following: histology consistent with borderline or non-epithelial carcinoma, surgical exploration at another institution, incomplete clinico-pathologic data, history of an additional primary tumour within the 5 years before or during the period after ovarian cancer diagnosis, and non-advanced-stage [International Federation of Gynecology and Obstetrics (FIGO) stages I and II] disease.

Forty-eight women whose primary presentation included bowel obstruction were identified. Bowel obstruction was diagnosed preoperatively on the basis of clinical findings such as nausea, vomiting, abdominal distension, absence of flatus/bowel movement and imaging such as distended bowel loops, presence of air–fluid levels, presence of transition points on computed tomography scan. The diagnosis was confirmed during surgery. Our institution’s policy regarding suspicion of bowel obstruction before surgery was to avoid the bowel preparation which we would normally carry out for all women scheduled for ovarian cancer surgery; in addition, the women were forewarned of the likelihood of ileostomy or colostomy. The majority of the women underwent small or large bowel resection with primary reanastomosis, a few women had a colostomy or ileostomy. Controls (women with no bowel obstruction at presentation) matched for age ±5 years, FIGO stage, grade of tumour, chemotherapy type, origin (ovarian versus fallopian versus peritoneal) and year of diagnosis ±1 year were also identified at a ratio of four control women to one woman presenting with bowel obstruction. Control women were selected without knowledge of outcome.

Once women were selected, a retrospective chart and pathology review were performed. Individual subject data were collected retrospectively from inpatient and ambulatory medical records. Surgical and pathology reports from the primary surgery were reviewed in all women. The following perioperative information was abstracted: date of surgery, surgical procedures performed, volume of residual disease. Surveillance information included adjuvant therapy administered, the date of last follow-up and the date of death/date of recurrence; preoperative serum CA125 levels and albumin were also recorded.

Statistical analysis

Continuous variables were evaluated by Student’s t test or Wilcoxon–Mann–Whitney U test for normally or non-normally distributed variables, respectively. Categorical variables were evaluated by chi-square test or Fisher’s exact test as appropriate for category size. Survival estimates were plotted using the Kaplan–Meier method. The log-rank test was used to quantify statistically these survival differences on univariate analysis. Length of survival was calculated from the date of initial surgery to the date of death; surviving women were counted at the date of last contact. Multivariate analyses were performed with a Cox proportional regression method. Covariates which had statistically significant prognostic values in univariate analyses were entered into a multivariate Cox proportional hazards model and nonsignficant factors were removed in a stepwise fashion. Hazard ratios are expressed relative to a baseline reference category. All statistical tests were two-sided and differences were considered statistically significant at < 0.05.

Statistical analyses including Kaplan–Meier curves were plotted using SPSS statistical software (version 16.0, SPSS, Inc, Chicago, IL, USA). All other data analyses were performed with Stata statistical software (version 9.2, Stata Corp LP, College Station, TX, USA).

Results

During the study period, 48 women with bowel obstruction and 192 controls were identified. The median follow-up was 19 months (range 1–90 months) for women with bowel obstruction at presentation and 20 months (range 1–139 months) for the controls.

Table 1 summarises the demographic and clinical characteristics of the study population. The criteria by which the groups had been matched: age at presentation, FIGO stage, grade of tumour, type of chemotherapy, origin of tumour (ovary versus fallopian tubes versus peritoneum) were similar between women with bowel obstruction and controls. In addition, no differences were seen between preoperative median serum CA125 and albumin levels. Serous histology was found in 87% and 75% of women with and without bowel obstruction, respectively. Although we were unable to ascertain ethnicity information in many women from both groups, among those women for whom this information was available, almost all were Caucasians. Information regarding performance status was not collected in most of these women. The majority of women in both groups received platinum and taxane-based chemotherapy after surgical debulking. Most women, 85% of the bowel obstruction group and 82% of the ‘no bowel obstruction group’ completed a full course of chemotherapy (at least six cycles).

Table 1.   Demographic and clinical characteristics of women with ovarian cancer
 CasesControlsP value
n = 48Range= 192Range 
  1. Values for continuous measurements are medians, unless otherwise specified.

Age (years)67(44–90)65(39–93)0.9
Stage
III [n (%)]33 (69) 132 (69) 0.9
IV [n (%)]15 (31) 60 (31) 0.9
Grade
1 [n (%)]1 (2) 4 (2) 0.8
2 [n (%)]29 (60) 116 (60)  
3 [n (%)]18 (38) 72 (38)  
Optimal cytoreduction [n (%)]36 (75) 150 (78) 0.7
Platinum/taxol chemotherapy [n (%)]45 (93) 178 (93) 0.9
Histology [n (%)]
Serous42 (87) 144 (75)  
Endometroid1 (2) 12 (6)  
Mucinous0 4 (2)  
Clear cell0 9 (5)  
Transitional cell1 (2) 5 (2)  
Mixed3 (6) 15 (8)  
Small cell0 1 (0.6)  
Carcinosarcoma1 (2) 1 (0.6)  
Undifferentiated0 1 (0.6)  
Origin [n (%)]
Ovarian44 (92) 176 (92) 0.8
Fallopian tube1 (2) 4 (2)  
Peritoneal3 (6) 12 (6)  
Bowel surgery [n (%)]48 (100) 81 (42) >0.001
CA125 (U/ml)680(20–8541)450(11–38 646)0.6
Albumin (g/dl)3(1.5–4.7)3.1(1–4.6)0.4

The site of the obstruction was in the small bowel in five women (10.4%), large bowel in 34 women (70.8%) and a combination of both small and large bowel in nine women (18.9%). Bowel surgery was performed in all 48 (100%) women who presented with bowel obstruction; in contrast, only 81 of the 192 (42%) women without bowel obstruction (< 0.001) underwent bowel surgery. The extent of cytoreductive effort (determined by the optimal cytoreduction rate) was similar between the two groups; 75% versus 78.1% (= 0.7).

Time to recurrence was significantly less in women who presented with bowel obstruction (Figure 1). The median progression-free survival (PFS) for women presenting with bowel obstruction was 19 months (95% CI, 11.9–22.6) compared with 21 months (95% CI, 16.2–25.5) in the matched cohort (= 0.01). Overall survival (OS) was also significantly decreased in women who presented with bowel obstruction (Figure 2), with a median OS of 22 months (95% CI, 18.9–33.4) compared with 35 months (95% CI, 27.1–40.1) in the matched control group (= 0.008).

Figure 1.

 Progression-free survival (PFS); all the women presenting with bowel obstruction versus all the control women, who presented without obstruction.

Figure 2.

 Overall survival (OS); women presenting with bowel obstruction versus all the control women, who presented without obstruction.

The median OS was 36 months (95% CI, 29.7–40.1) for all women who underwent optimal cytoreduction compared with 19 months (95% CI, 15–26.7) in women whose cytoreduction was suboptimal (= 0.0001). When women who underwent optimal cytoreductive surgery were analysed, the median PFS was 20 months (95% CI, 12–28.43) among women presenting with bowel obstruction and 24 months (95% CI, 16.7–30.3) among controls (P = 0.01) (Figure 3). The median OS was 23 months (95% CI, 18.9–37.7) in women presenting with bowel obstruction versus 38 months (95% CI, 30.9–47.6) in controls (P = 0.009) (Figure 4).

Figure 3.

 Progression-free survival (PFS); women who underwent optimal cytoreduction.

Figure 4.

 Overall survival (OS); women who underwent optimal cytoreduction.

The median follow-up was shorter than the median OS, to ensure that our findings remain valid at the time of median follow-up (19 months for women presenting with bowel obstruction and 20 months for controls), we calculated OS at 20 months. The survival rate was 58% for women presenting with bowel obstruction and 67% for controls (P = 0.03).

In the multivariate Cox regression model, bowel obstruction was an independent predictor of OS (hazard ratio 1.55; 95% CI, 1.05–2.3, P = 0.009) as shown in Table 2. Other significant factors studied included stage (P = 0.01), optimal cytoreduction versus suboptimal cytoreduction (P < 0.001), platinum and taxane-based chemotherapy versus other agents (P = 0.003) and age (P < 0.001). Tumour grade was not significant and so was excluded from the model.

Table 2.   Multivariate Cox proportional hazards model for recurrence and overall mortality
 HR95% CIP value
Presentation: bowel obstruction versus no bowel obstruction1.551.05–2.30.009
Stage: IV versus III1.51.1–2.20.01
Residual: greater than 1 cm versus 1 cm or less1.61.1–2.3<0.0001
Chemotherapy: platinum and taxol versus an alternative agent1.71.1–3.20.003
Age (continuous)1.021–1.03<0.0001

Discussion

Ovarian cancer is one of the most chemo-responsive solid tumours with objective response rates ranging from 41% to 73% in prospective phase III trials. 18–20 Despite this impressive chemo-responsiveness, the majority (73–76%) of women with ovarian cancer will experience disease recurrence following initial treatment and remission.19 Previously reported adverse prognostic factors in women with ovarian cancer include stage, age, performance status, tumour histology and residual tumour volume following cytoreduction.17,21,22

Although ovarian cancer sometimes metastasises to extra-abdominal sites such as lung,23 brain24–26 and bone27,28 the most important co-morbidity of ovarian cancer relates to bowel compromise.29–32 It is therefore not surprising that bowel obstruction and or perforation is the cause of death in the majority of women with recurrent ovarian cancer.13,33,34

Although, bowel obstruction is seen in 8–30% of women at the time of initial presentation for ovarian cancer,11,12 to our knowledge, the impact of this mode of presentation on prognosis following standard therapy has never been evaluated. This study suggests that bowel obstruction as part of primary presentation has an adverse effect on outcome. We thought of several reasons that might have accounted for the less favourable outcome in this subgroup of women and these are discussed.

First, women with bowel obstruction at presentation may require a more extensive cytoreductive effort including bowel resection in order to achieve an optimal status at the end of primary surgery. In our study, the rate of optimal cytoreduction was the same among the two groups of women (75% versus 78%, with and without bowel obstruction on presentation, respectively), so one cannot attribute the survival difference to suboptimal status after primary debulking. One previous study showed an improved disease-free survival in women who required bowel resection to achieve an optimal status. The women in this study who presented with bowel obstruction were compared with the suboptimally debulked control women.11 We contend that the reported improvement in disease-free survival in this particular study is attributable to the extent of cytoreductive surgery and not to bowel resection.

Second, it is possible that women who present with bowel obstruction have different tumour biology to those that do not have an obstruction at presentation and it is this difference in tumour biology that accounts for the survival difference. During our study development, we controlled for known surrogates of tumour biology (stage of tumour, tumour grade, site of origin and histological type). Regardless of this fact, a survival difference was noted between women from the two groups. Clearly these known surrogates do not identify all the molecular and genetic characteristics of these tumours that may explain the survival differences (Table 1).

Third, we examined the type of adjuvant chemotherapy used for the women, the proportion of women given platinum and taxane-based chemotherapy was strikingly similar in both groups (Table 1).

Last, we hypothesise that when bowel obstruction is present at the time of primary presentation of ovarian cancer, it leads to micro-metastasis in the bowel mesenteric lymph nodes which may eventually cause a shorter PFS and OS. This reasoning stemmed from the fact that this subgroup of women had a shorter PFS and OS compared with the controls regardless of optimal cytoreduction rate. Salani et al. did show improved survival in a subgroup of women undergoing ovarian cancer debulking who had an adequate mesenteric resection along with the resected segment of sigmoid colon. Micrometastasis was frequently seen in these women’s sigmoid mesocolon lymph nodes.35

The median PFS of 19 months and the median OS of 22 months are rather close. We are not entirely certain why this happened; however, when ovarian cancer recurs, it is invariably fatal. The commonest cause of death from ovarian cancer is bowel complication so it is possible that the cohort of women in this study had bowel involvement quickly after recurrence, leading to rapid fatality.

Other well-known prognostic factors in ovarian cancer — age, stage at presentation and size of residual tumour — were all found to be significant in the Cox regression analysis (Table 2). Of particular interest is the persistent influence of bowel obstruction at presentation in our subgroup analysis. When we limited our survival analysis to only optimally cytoreduced women, those who presented initially with bowel obstruction still had a shorter PFS and OS (Figures 3 and 4, respectively).

This is a small, single institution, retrospective study. The limitations imposed by these attributes have to be borne in mind when interpreting or using the findings of this study.

However, our findings are interesting and add to the array of prognostic factors available to physicians evaluating women with ovarian cancer. Furthermore, future prospective studies should pay careful attention to the subgroup of women with ovarian cancer with bowel obstruction at the time of presentation as different strategies may be required to improve their outcome. Given its potential influence on PFS and OS, this characteristic should be used to stratify women when interpreting data and should be equally distributed in randomisations.

We conclude that bowel obstruction at the time of primary presentation of ovarian cancer is an indicator of a less favourable outcome following therapy. This information may be helpful in counselling and guiding management.

Disclosure of interests

None of the authors of this manuscript has any conflict of interest to declare.

Contribution to authorship

All the authors listed on this manuscript have contributed to its development and preparation. JAR-H and ABO came up with the idea. The concept was discussed with AKG, MC, NSH, EK and IAA. IAA and EK helped with data collection. AKG, NSH and MC helped in manuscript planning and draft. All the authors participated in developing the manuscript.

Details of ethics approval

Our institutional review board (IRB) approval was requested and granted to study malignant bowel obstruction in women with ovarian cancer. This study is one of two publications resulting from this IRB approval. The protocol approval number is 2007-P-001900/2; MGH (Massachusetts General Hospital).

Funding

No funding was solicited or required for this study.

Acknowledgement

None.

Ancillary