Annals of Surgical Oncology

, Volume 16, Issue 5, pp 1324–1330

Surgery for Recurrent Ovarian Cancer: Role of Peritoneal Carcinomatosis: Exploratory Analysis of the DESKTOP I Trial About Risk Factors, Surgical Implications, and Prognostic Value of Peritoneal Carcinomatosis


    • Department of Gynecology and Gynecologic OncologyHSK, Dr. Horst Schmidt Klinik
  • M. Hahmann
    • Coordinating Center for Clinical TrialsMarburg University
  • H. J. Lueck
    • Department of Gynecology and ObstetricsMedizinische Hochschule Hannover
  • M. Poelcher
    • Department of Gynecology and ObstetricsBonn University
  • P. Wimberger
    • Department of Gynecology and ObstetricsEssen University
  • O. Ortmann
    • Department of Gynecology and ObstetricsCaritas St. Josef Hospital
  • U. Canzler
    • Department of Gynecology and ObstetricsDresden University
  • B. Richter
    • Department of Gynecology and ObstetricsKlinikum Radebeul
  • U. Wagner
    • Department of Gynecology and ObstetricsMarburg University
  • A. Hasenburg
    • Department of Gynecology and ObstetricsFreiburg University
  • A. Burges
    • Department of Gynecology and ObstetricsLMU Munich
  • S. Loibl
    • Department of Gynecology and ObstetricsFrankfurt University
  • W. Meier
    • Department of Gynecology and ObstetricsEVK Düsseldorf
  • J. Huober
    • Department of Gynecology and ObstetricsTübingen University
  • D. Fink
    • Department of Gynecology and ObstetricsZurich University
  • W. Schroeder
    • Department of Gynecology and ObstetricsKlinikum Bremen Mitte
  • K. Muenstedt
    • Department of Gynecology and ObstetricsGiessen University
  • B. Schmalfeldt
    • Department of Gynecology and ObstetricsMunchen rdI
  • G. Emons
    • Department of Gynecology and ObstetricsGoettingen University
  • A. du Bois
    • Department of Gynecology and Gynecologic OncologyHSK, Dr. Horst Schmidt Klinik
Gynecologic Oncology

DOI: 10.1245/s10434-009-0357-0

Cite this article as:
Harter, P., Hahmann, M., Lueck, H.J. et al. Ann Surg Oncol (2009) 16: 1324. doi:10.1245/s10434-009-0357-0



Almost all retrospective trials pointed out that a benefit of surgery for recurrent ovarian cancer may be limited to patients in whom a macroscopic complete resection could be achieved. Peritoneal carcinomatosis has been reported to be either a negative predictor for resectability or a negative prognostic factor, or both. The role of peritoneal carcinomatosis in a multicenter trial was investigated.


Exploratory analysis of the DESKTOP I trial (multicenter trial of patients undergoing surgery for recurrent ovarian cancer, 2000 to 2003).


A total of 125 patients (50%) who underwent surgery for recurrent ovarian cancer had peritoneal carcinomatosis. Univariate analyses showed worse overall survival for patients with peritoneal carcinomatosis compared with patients without carcinomatosis (< .0001). Patients with and without peritoneal carcinomatosis had a complete resection rate of 26% and 74%, respectively (< .0001). This corresponded with the observation that patients with complete resection had a better prognosis than those with minimal residual disease of 1 to 5 mm, which commonly reflects peritoneal carcinomatosis (P = .0002). However, patients who underwent complete resection, despite peritoneal carcinomatosis, had a 2-year survival rate of 77%, which was similar to the 2-year survival rate of patients with completely debulked disease who did not have peritoneal carcinomatosis (81%) (P = .96). Analysis of prognostic factors did not show any independent effect of peritoneal carcinomatosis on survival in patients who underwent complete resection.


Peritoneal carcinomatosis was a negative predictor for complete resection but had no effect on prognosis if complete resection could be achieved. Improving surgical skills might be one step to increase the proportion of patients who might benefit from surgery for recurrent disease.

Most series about surgery for recurrent ovarian cancer reported complete debulking as independent prognostic factor for further survival after surgery.18 In the largest series about surgery for recurrent ovarian cancer, the AGO DESKTOP I trial, complete resection was the strongest prognostic factor for survival.9 Many series, including all larger series, reported a survival benefit only in patients who had undergone complete debulking. The AGO DESKTOP I trial showed a median overall survival of 45 months in patients with completely debulked disease compared with 19 months in patients with incomplete resection of recurrent disease (hazard ratio [HR], 3.71; 95% confidence interval [95% CI], 2.27–6.05; < .0001). There were no differences between patients with so-called optimal debulking (1 to 10 mm) or residual disease of > 1 cm in size after surgery (HR .84; 95% CI, .51–1.40; P = .502).

A recently published series by Chi et al. also showed a benefit in patients with limited residual disease up to 5 mm after surgery.10 Residual disease of ≤ 5 mm usually presents peritoneal carcinomatosis. Peritoneal carcinomatosis is presently described in approximately 35% to 43% of patients with a primary diagnosis of ovarian cancer, and up to 85% in stage IIIC patients.1113 In a recently published survey of the members of the Society of Gynecologic Oncology, a “prohibitive number of implants” was ranked as the number 1 reason for inoperability (71.9% of inoperable patients), followed by “disease in hazardous locations” (20.7%).14

Therefore, the following questions remained unanswered: (1) What are the risk factors for peritoneal carcinomatosis at relapse? (2) How often is complete resection possible in patients with peritoneal carcinomatosis? (3) How is a successful surgery in patients with peritoneal carcinomatosis defined? (4) Are there any prognostic differences between patients with and without peritoneal carcinomatosis? Therefore, we analyzed the database of the AGO DESKTOP I trial with special emphasis on peritoneal carcinomatosis.

Materials and Methods

Twenty-five centers from Germany and Switzerland, all members of the AGO Ovarian Committee and/or the study coordinating group of the AGO-OVAR, enrolled all patients with recurrent invasive epithelial ovarian cancer who received cytoreductive surgery between January 2000 and December 2003 in this retrospective trial. Data were extracted from patients’ records and documented according to a standardized database. No personal data were collected, and only treating physicians could identify patients. All data were checked for plausibility and completeness by two authors (P.H. and A.dB.), and queries were answered by phone and evaluation of surgical and pathological reports, which were cleared of personal data beforehand.

Patients with nonepithelial ovarian cancer or tumors of low malignant potential were excluded. Surgeries with symptom-oriented and strictly palliative purposes or operations within the context of primary surgery (e.g., interval debulking, second-look surgery) were excluded. Furthermore, patients with missing data about peritoneal carcinomatosis were excluded for this analysis. Ascites was defined as present if there was > 500 mL of fluid, and peritoneal carcinomatosis was defined as present according to the surgical report. Further details have been previously published.9

Data were analyzed by descriptive statistics. Frequency counts and percentages were used to describe categorical variables, and mean and range were used for continuous variables. Logistic regression models were used to identify associations between different risk factors and the probability of favorable surgical outcome, and Cox regression models and Kaplan-Meier estimates were used to explore the impact of different covariates on overall survival. Statistical significance was defined as a P value of < .05, and both-sided tests were applied. Statistical analysis was performed with SAS software, version 9.1 (SAS Institute, Cary, NC).


The DESKTOP I trial included 267 patients, of whom 17 with missing data about intraoperative findings of peritoneal carcinomatosis were excluded for this analysis. The remaining 250 patients represented the study collective. These patients had a median age of 60 years (range, 24–84 years), and most of them had good performance status (78.4% with Eastern Cooperative Oncology Group score 0 or 1). The patient and tumor characteristics, including prior surgery and chemotherapy, of all patients of the DESKTOP I trial have already been published.9 A total of 125 (50%) of 250 patients had peritoneal carcinomatosis diagnosed during surgery. Patients with and without peritoneal carcinomatosis had comparable characteristics but different tumor characteristics. Patients with peritoneal carcinomatosis at relapse were more likely to have had advanced International Federation of Gynecology and Obstetrics (FIGO) stage at primary diagnosis (P = .003), lower rates of complete resection at first surgery (P = .0007), shorter treatment-free interval (P = .0011), and more ascites (< .0001). In multivariate analysis, only ascites (P = .0001) and treatment-free interval (P = .0329) remained statistically significant as risk factors for peritoneal carcinomatosis (Tables 1 and 2).
Table 1

Patient and tumor characteristics of patients with and without peritoneal carcinomatosis (PC)


PC (n = 125)

Non-PC (n = 125)

Median age (y) (range)

59 (27–81)

61 (24–84)

ECOG performance status


53 (42.4%)

62 (49.6%)


72 (57.6%)

63 (50.4%)

FIGO stage


28 (22.4%)

50 (40.0%)


97 (77.6%)

75 (60.0%)

Complete resection first surgery


45 (36.0%)

72 (57.6%)


80 (64.0%)

53 (42.4%)

Treatment-free interval (mo)


22 (17.6%)

12 (9.6%)


39 (31.2%)

21 (16.8%)


64 (51.2%)

92 (73.6%)



96 (76.8%)

121 (96.8%)


29 (23.2%)

2 (3.2%)

ECOG Eastern Cooperative Oncology Group; FIGO International Federation of Gynecology and Obstetrics

Table 2

Univariate and multivariate analysis of predictive factors for peritoneal carcinomatosis



Standard error

OR (95% CI)

P value


    FIGO stage (I/II vs. III/IV)



2.309 (1.330–4.011)


    Complete resection at first surgery (yes vs. no)



2.415 (1.451–4.018)


    Treatment-free interval (<6 vs. 6–12 vs. >12 mo)



1.835 (1.273–2.643)


    Ascites (no vs. yes)



9.138 (3.106–26.886)


    ECOG (0 vs. >0)



1.337 (.812–2.202)



    Ascites (no vs. yes)



8.612 (2.844–26.074)


    FIGO stage (I/II vs. III/IV)



1.75 (.919–3.332)


    Complete resection at first surgery (yes vs. no)



1.792 (.991–3.241)


    Treatment-free interval (<6 vs. 6–12 vs. >12 mo)



.658 (.447–.967)


FIGO International Federation of Gynecology and Obstetrics; ECOG Eastern Cooperative Oncology Group

Complete resection was achieved in 50% of all patients, irrespective of intraoperative findings, but if peritoneal carcinomatosis was present, complete resection was possible in only 26% of patients, compared with 74% of patients without peritoneal carcinomatosis (< .0001). Debulking to 1 to 5 mm in patients with peritoneal carcinomatosis was achieved in additional 23%, to 6 to 10 mm in 14% and to > 10 mm in 37%, respectively. The corresponding percentages in patients without peritoneal carcinomatosis were 9% (1 to 5 mm), 7% (6 to 10 mm), and 10% (> 10 mm).

Patients with intraoperative findings of peritoneal carcinomatosis had a worse prognosis than patients without peritoneal carcinomatosis (P < .0001; HR, 2.724; 95% CI, 1.675–4.431) (Fig. 1). The median overall survival was 19.9 months for all patients and 45.3 months for patients with and 35.2 months for patients without peritoneal carcinomatosis, respectively. Patients with peritoneal carcinomatosis showed statistically significant differences in overall survival if their disease could be completely debulked, compared with patients with any postoperative residual disease. However, no differences could be shown between minimal residual disease (1 to 5 mm), optimal debulking (6 to 10 mm), and suboptimal debulking (> 10 mm) (Fig. 2).
Fig. 1

Survival in patients with (PC) and without (NPC) peritoneal carcinomatosis
Fig. 2

Survival of patients with peritoneal carcinomatosis (PC) after cytoreductive surgery and residual disease

Patients who had undergone complete resection despite peritoneal carcinomatosis had a 2-year survival rate of 77%, which was similar to the 2-year survival rate of patients without peritoneal carcinomatosis whose disease was completely debulked (81%) (HR 1.028, 95% CI, .394–2.679; P = .96) (Fig. 3). The following factors were investigated by univariate analysis of completely debulked disease: age (P = .1202), performance status (P = .6268), residual disease after primary surgery (P = .0128), postoperative chemotherapy (P = .0689), ascites (≤ .0001), peritoneal carcinomatosis (P = .9552), and treatment-free interval (P = .0022). In addition, multivariate analysis including residual disease after primary surgery and treatment-free interval in patients with completely debulked disease was performed, but only residual disease after primary surgery remained statistically significant (Tables 3 and 4). The presence of ascites was excluded from multivariate analysis because only three patients with ascites were observed. The rate of relaparotomies for complications was 3.6% (nine patients). Four patients had thrombosis (2%), but only one patient experienced pulmonary embolism. Eight patients (3.2%) had secondary wound healing. An additional 25 patients (10%) had other complications, but no patient died within 30 days after surgery.
Fig. 3

Comparison of patients with (PC) and without (NPC) peritoneal carcinomatosis whose disease underwent complete dubulking

Table 3

Univariate analysis of prognostic factors in patients with complete debulked disease


n (%)

P value

Hazard ratio (95% confidence interval)

Age (y)


56 (45%)




69 (55%)



ECOG performance status


73 (58%)




52 (42%)



Complete resection 1st surgery


71 (57%)




54 (43%)



Treatment-free interval (mo)


16 (13%)




24 (19%)




85 (68%)




120 (98%)




3 (2%)



Peritoneal carcinomatosis


33 (26%)




92 (74%)



Platinum-based chemotherapy after second surgery


55 (44%)




70 (56%)



ECOG Eastern Cooperative Oncology Group

Table 4

Multivariate analysis of prognostic factors in patients with completely debulked diseasea



Standard error

Hazard ratio (95% confidence interval)

P value

Complete resection first surgery (yes vs. no)



3.176 (1.163–8.669)


Treatment-free interval (< 6 vs. 6–12 vs. > 12 months)



1.359 (.816–2.265)


aThe presence of ascites was excluded in multivariate analysis because of the small sample size


There is still no level II evidence for surgery for recurrent ovarian cancer, and heterogenous selection bias in all retrospective series did not help us learn whether a favorable course of disease after secondary surgery reflects tumor biology alone. However, many authors showed consistently an excellent prognosis of up to 100 months’ survival in patients with complete resection, far exceeding survival durations observed in chemotherapy-only series.8,10,1517 The prognostic role of the maximum size of residual tumor has been reported in numerous studies in primary surgery.18,19 Complete resection without macroscopic residual tumor has more recently been defined as goal of primary surgery.20 However, so-called optimal cytoreduction to residual tumors up to 1 cm in size still make a difference compared with larger residuals.21 This effect did not seem to account for surgery in relapsed ovarian cancer. Most authors showed a survival benefit after surgery only for patients who had undergone complete resection.1,5,8,22 Only one larger series, published by Chi et al., proposed a survival benefit after surgery with minimal residual disease (< .5 mm).10 However, in their series, 41% underwent complete resection, and only 11% of the 153 patients had residual disease between 1 and 5 mm in size. The Kaplan-Meier curve showed a worse prognosis of this subgroup compared with patients whose disease was completely debulked. The small number of patients with minimal residual disease in this series might explain the nonsignificant results.

This study reflects the difficulties in small retrospective single-center series with immanent selection biases, leading to a great variety of reported complete resection rates—between 17% and 81%.1,7 The largest multicenter trial (DESKTOP I) showed a complete resection rate of 50% for the whole study population and could not detect any beneficial impact of small residual disease. This is comparable to another actual published series of 217 patients from the Radium Hospital, with a complete debulking rate of 49%.23 Unfortunately, the reasons responsible for leaving any residual tumor were not documented in the DESKTOP I series. However, the presence of peritoneal carcinomatosis might have played a role because the intraoperative finding of peritoneal carcinomatosis was a statistically significant negative predictor for complete debulking.

These findings correspond with the reports of Eisenkop et al. and Chi et al., who also investigated peritoneal carcinomatosis as negative factor for complete resection.1,10 In addition, patients with peritoneal carcinomatosis had worse prognosis compared with patients without peritoneal carcinomatosis. However, it was unclear whether peritoneal carcinomatosis itself presents a negative tumor biologic factor or indicates only a technical burden associated with an inferior rate of complete resection. The latter hypothesis was supported by our findings. In our series, complete resection was possible in 24% of patients with peritoneal carcinomatosis only, compared with 76% of patients without this spread. However, no prognostic differences could be found for patients with or without peritoneal carcinomatosis if complete resection of recurrent disease could be achieved. Eisenkop et al. have already found that cytoreduction to a visibly disease-free outcome had a more important influence on survival than the extent of metastatic disease present before surgery in primary ovarian cancer.24 Furthermore, this observation contradicts a specific tumor biologic prognosis factor associated with peritoneal carcinomatosis. Technical problems might be overcome by development of better techniques and surgical skills.

The subgroup of patients with peritoneal carcinomatosis contributes substantially to the whole cohort of patients with recurrent ovarian cancer.25 In these patients, deperitonealization might be beneficial if complete resection is achieved. This increment of surgical radicality has to be balanced with potential benefit and treatment burden. Currently, we do not have very well-established criteria to select those patients who will benefit from this procedure and those who will not. One primary objectives of the DESKTOP I trial was to build a hypothesis for a score predicting complete resection (AGO score). This model score consisted of three factors: good performance status ECOG 0, complete resection at primary surgery (or alternatively, early FIGO stage at first diagnosis), and absence of ascites. In patients in whom all of these three items are present, complete resection was possible in 79%.

Other models that include peritoneal carcinomatosis as variable have been suggested.10 A similar calculation of a predictive score including this variable has been performed in the DESKTOP I trial. This model was not reported, although it fitted well with a potential prediction of resectability, because we found it to be useless: no reliable information about peritoneal carcinomatosis is available before surgery. Computed tomographic scan and/or ultrasound are not suitable for accurate diagnosis of peritoneal carcinomatosis.26 Some authors suggested open laparoscopy for estimation of resectability.8 However, the rate of complete debulking was 81% after open laparoscopy and was similar to the results that used the preoperative AGO score. The prospective evaluation of this score has been the primary endpoint of the recently closed DESKTOP II trial, and results will probably be available soon.

Further studies should especially focus on peritoneal carcinomatosis and ways to overcome nonresectability. In this context, the present study has its limitations. The primary objectives of the DESKTOP I trial were the identification of the appropriate surgical aim and the creation of a predictive score for a successful surgery. In this context, peritoneal carcinomatosis was not quantified and was only classified as absent or present. This ignored the fact that from a surgeon’s view, diffuse visceral carcinomatosis is not comparable with strictly parietal carcinomatosis on the peritoneal surface of the paracolic gutters, pelvis, or diaphragm. Exact data about the pattern of carcinomatosis in recurrent ovarian cancer are not yet available but should be collected in future trials. The currently planned prospectively randomized DESKTOP III trial, which will evaluate the role of cytoreduction in recurrent ovarian cancer, will document the incidence, pattern of spread, and surgical techniques to resect peritoneal carcinomatosis. This study will also evaluate the treatment burden and its implication on quality of life. The latter might help balance the benefit of more radical surgical procedures, like stripping of the diaphragm or other upper abdominal surgeries, which have been described to be beneficial for cytoreduction and of overall patient benefit.2729


Further members of the AGO OVAR and/or AGO Ovarian committee who contributed to this study are: J. Pfisterer (Kiel), K. Wollschlaeger (Magdeburg), H. G. Meerpohl (Karlsruhe), G. P. Breitbach (Neunkirchen), B. Tanner (Berlin), J. Sehouli (Berlin), and V. Heil (Ulm). We thank S. Eichner, A. Krüger, M. Schulze, C. Ackermann, and G. Elser (AGO-OVAR study office) for data management and technical support, and J. Rau, A. Reuss, and C. Schade-Brittinger (KKS Marburg) for statistical support.

Copyright information

© Society of Surgical Oncology 2009