Objective
The aim of this study was to compare carcinomatosis scores, and to determine their relevance to predict resectability, morbidity, and outcome.
Study Design
From 2005-2008, 61 patients underwent surgery for ovarian cancer. We compared International Federation Gynecology and Obstetrics (FIGO), peritoneal cancer index, Eisenkop, Aletti, Fagotti, and Fagotti-modified scores.
Results
There was a strong correlation between the different scores. In predicting resectability, Fagotti-modified and peritoneal cancer index outperformed other scores. We demonstrated a strong association between the occurrence of postoperative complications and Aletti, peritoneal cancer index, and Eisenkop scores ( P < .0001). For progression-free survival, we observed significant differences among FIGO, peritoneal cancer index, Eisenkop, Fagotti-modified, and Aletti stages ( P < .05). For stage III/IV patients, only Aletti score remains significant to predict resectability. This suggests that complete respectability is more related to the surgical effort than to the extent of the disease.
Conclusion
Alternative ranking systems provide additional information over FIGO for complete resectability, complications, and survival.
Ovarian cancers are usually diagnosed at an advanced stage, with massive, widespread intraabdominal disease. Multiple series report that the completeness of the primary cytoreductive operations independently influences survival for these patients. Median and 5-year survivals after removal of all visible disease (complete cytoreduction) are reported to exceed survival resulting from suboptimal procedures (residual disease ≤ 1 cm). The extent of the disease before surgery partly determines the ability to perform a complete cytoreduction. The International Federation Gynecology and Obstetrics (FIGO) staging system, introduced in 1964, is used for the evaluation of the peritoneal spread, according to operative findings. However, for advanced ovarian cancer (stages IIIC and IV), there is a great disparity among patients of similar stage. To quantify with more precision the intraabdominal extent of the disease, a number of numerical ranking systems have been proposed. The first one was the peritoneal cancer index (PCI), used to describe the peritoneal spread in intraperitoneal and pelvic malignant tumors. Its use for ovarian cancer has also been reported. Eisenkop et al developed a score specific for ovarian cancer that is based on the spread of peritoneal and retroperitoneal disease. Recently, Fagotti et al reported a laparoscopy-based score for predicting surgical resectability at the beginning of the surgery and Brun et al suggested a Fagotti-modified score. Finally, Aletti et al reported a different score for ovarian cancer, which reflects the surgical complexity adjusted for the extent of the surgery. To our knowledge, no head-to-head comparisons of these scores have been reported in the literature.
The aim of our study was therefore to compare these 6 scores (FIGO, PCI, Eisenkop, Fagotti, Fagotti-modified, and Aletti) to determine their relevance to predict resectability, surgical morbidity, and outcome.
Materials and Methods
Between 2005 and 2008, 61 patients underwent surgery for ovarian cancer in Tenon hospital, Paris, France. Individual records for all patients were reviewed and analyzed. Patient and tumor characteristics, intraoperative findings, postoperative courses and outcomes were analyzed. For each patient, we determined the FIGO stage, the PCI, and the Eisenkop score. Residual disease was reported as complete (no residual disease), or not. Intraoperative units of blood transfusion, operative time, and length of hospitalization (including length of time in the intensive care unit) were all recorded. The Fagotti and Fagotti-modified scores were calculated when a laparoscopy was performed before the debulking surgery. The Fagotti and the Aletti scores were evaluated retrospectively on operative reports (score published in 2006 and 2007). The percentages of complete resections and complications were calculated according to different stages (as reported in original reports) for each score.
We calculated the correlation between scores in a pairwise fashion, and established a receiver operating characteristic (ROC) curve to perform concordance index analyses. We assessed the ability of the models to identify patients who were most likely to have complete surgical resection or postoperative complications. We also determined the relevance of each score in predicting outcome.
Score description is detailed in the Appendix. Thresholds that were used had previously been reported as relevant.
Statistical analysis
Data were analyzed using the Fisher’s exact test, the χ 2 test, the Student t test, and the nonparametric Mann-Whitney and Wilcoxon tests, if necessary. The survival was analyzed using Kaplan-Meier survival curves, and the log-rank test (univariate) was used to analyze the relationship between scores and survival. The progression-free survival was measured from the initial treatment to the date of relapse (defined by Response Evaluation Criteria in Solid Tumors [RECIST] and/or CA125 criteria). The model performance was quantified with respect to discrimination. Discrimination (ie, whether the relative ranking of individual predictions is in the correct order) was quantified either with the area under the ROC curve (AUC) or with the concordance index, which is similar to the AUC, but is appropriate for censored data. ROC curve analysis was used to assess the ability of the model to identify patients who were most likely to have complete surgical resection or postoperative complication. The concordance index is the probability that given 2 randomly selected patients, the patient with the worse outcome will in fact have a worse outcome prediction. The concordance index ranges from 0-1, with 1 indicating perfect concordance, 0.5 indicating no better concordance than chance, and 0 indicating perfect discordance.
Differences were considered significant at a level of P < .05.
Results
Sixty-one patients were included in this study. The median age of the patients was 57.2 years. According to FIGO classification, 13% of patients were stage I (IA = 5, IC = 3), 8% stage II (IIA = 2, IIB = 3), 71% stage III (all IIIC, n = 43), and 8% stage IV (n = 5). Sixty-one percent had a serous histologic type and 88% had a tumor grade of 2 or 3. The mean preoperative CA125 level was 1526 UI/mL. Twenty-eight percent of the patients received neoadjuvant chemotherapy. For these patients, scores at interval surgery (and not at diagnostic laparoscopy) were taken into account.
Surgical procedures ( Table 1 ) included hysterectomy for 90% of patients, omentectomy for 95%, intestinal resection for 60%, and pelvic or paraaortic lymphadenectomy for 75% of patients. Peritoneal procedures were performed on 57% of patients. Complete resection (no residual disease) was achieved in 80% of the cases. During surgery, 41% of the patients required blood transfusion, with a mean estimated blood loss of 955 mL. The mean operative time was 425 minutes. The mean length of hospitalization was 1.8 days in the intensive care unit and 12.7 days for the whole hospitalization. Postoperative complications occurred in 54% (n = 33) of the patients, including fever (52% of complications), digestive tract fistula (6%), or symptomatic lymphocyst (24%). Eighteen percent of patients were rehospitalized (n = 11: 3 lymphocyst drainage, 2 pleural puncture, 4 surgical complications, and 2 complications of chemotherapy), and 23% had iterative surgery or interventional radiology (lymphocyst drainage).
| Characteristics | n = 61 | % |
|---|---|---|
| Median age, y | 57 | |
| Neoadjuvant chemotherapy | 17 | 28 |
| Standard surgical procedures | ||
| Hysterectomy | 55 | 90 |
| Unilateral or bilateral salpingo-oophorectomy | 59 | 97 |
| Omentectomy | 58 | 95 |
| Intestinal resection | 37 | 60 |
| Appendectomy | 30 | 49 |
| Pelvic lymphadenectomy | 46 | 75 |
| Paraaortic lymphadenectomy | 45 | 74 |
| Extensive surgery | ||
| Splenectomy | 6 | 10 |
| Cholecystectomy | 5 | 8 |
| Lesser omentum resection | 14 | 23 |
| Peritoneal surgery | 35 | 57 |
| Diaphragmatic surgery | ||
| Right | 27 | 44 |
| Left | 15 | 25 |
| Douglassectomy | 31 | 51 |
| Prevesical peritoneum | 26 | 43 |
| Complete resection | 49 | 80 |
| Mean of each score (range) | ||
| PCI (0-39) | 16 | |
| Eisenkop score (0-15) | 7 | |
| Aletti score (0-18) | 7.6 | |
| Fagotti (0-14) | 5.6 | |
| Fagotti-modified (0-8) | 2.1 |
We first studied the correlations between the scoring systems for all FIGO stages. Using a correlation test ( Table 2 ), we found a strong correlation between the Aletti and Eisenkop scores (correlation coefficient, 0.8; P < .0001) and the Aletti and PCI scores (correlation coefficient, 0.76; P < .0001). Also, the PCI and Fagotti (correlation coefficient, 0.84; P < .0001) and the PCI and Fagotti-modified scores (correlation coefficient, 0.8; P < .0001) were strongly correlated. Moreover, the Eisenkop score was strongly correlated with the PCI (correlation coefficient, 0.94; P < .0001) and with the Fagotti score (correlation coefficient = 0.81; P < .0001). The Fagotti and Fagotti-modified score were also strongly correlated (correlation coefficient, 0.8; P < .0001). Interestingly, the FIGO score was only fairly correlated with the other scoring systems (correlation coefficient range from 0.44-0.64). We analyzed patients by subgroups (patients who received neoadjuvant chemotherapy or initial surgery), we found more similar results than in the whole group: PCI and Fagotti-modified scores were the more relevant ( Appendix Table 5 ). Pairwise analysis of correlation between scores was significant (paired Wilcoxon test, P = .02), suggesting that the scores can be used independently of the therapeutic sequence.
| Variable | Aletti | Eisenkop | PCI | Fagotti | FIGO | Fagotti-modified |
|---|---|---|---|---|---|---|
| Aletti | 1 | 0.8 a | 0.76 a | 0.59 a | 0.44 b | 0.41 c |
| Eisenkop | — | 1 | 0.94 a | 0.81 a | 0.64 a | 0.78 a |
| PCI | — | — | 1 | 0.84 a | 0.59 a | 0.8 a |
| Fagotti | — | — | — | 1 | 0.61 a | 0.8 a |
| FIGO | — | — | — | — | 1 | 0.6 a |
| Fagotti-modified | — | — | — | — | — | 1 |
We next studied the discriminatory power of each score in terms of complete resection for the whole population ( Figure 1 ). The best AUC, 0.76, was obtained with the Fagotti-modified score. This result was expected because this score was developed using this series. PCI score also had relevant AUCs (0.69; P = .02). The other scores had AUCs between 0.6 and 0.7, without a significant difference. Aletti score was not relevant for discriminating patients for complete resection (AUC, 0.33; P = .9). This was expected because this score was developed for predicting surgical complications.
We next analyzed the rate of complete resections according to the ordinal scales of scores ( Table 3 ). The mean PCI score was 16. PCI scores <10 were related to more complete resections (92% vs 71%; P = .04). We found similar results with Aletti score with more complete resections ( P = .05) being observed with an increase in the score. For the other scoring methods, there was no significant difference among the different stages for complete resection.
| Variable | n | % | % complete resection | P | % complications | P |
|---|---|---|---|---|---|---|
| ALL FIGO STAGES | ||||||
| PCI | ||||||
| PCI <10 | 26 | 43 | 92 | 31 | ||
| PCI >10 | 35 | 57 | 71 | .04 | 71 | .002 |
| Eisenkop score | ||||||
| Eisenkop <6 | 27 | 44 | 89 | 30 | ||
| Eisenkop 6-10 | 17 | 28 | 65 | 53 | ||
| Eisenkop >10 | 17 | 28 | 82 | .141 | 94 | < .0001 |
| Aletti score | ||||||
| Low (≤3) | 8 | 13 | 50 | 0 | ||
| Intermediate (4-7) | 26 | 43 | 80 | 42 | ||
| High (≥8) | 27 | 44 | 89 | .05 | 81 | < .0001 |
| Fagotti | ||||||
| Fagotti <8 | 22 | 58 | 86 | 36 | ||
| Fagotti ≥8 | 16 | 42 | 81 | .68 | 81 | .006 |
| Fagotti-modified | ||||||
| <4 | 23 | 61 | 91 | 43 | ||
| ≥4 | 15 | 39 | 73 | .19 | 73 | .07 |
| FIGO stage | ||||||
| I | 8 | 13 | 100 | 0 | ||
| II | 5 | 8 | 100 | 60 | ||
| III | 43 | 71 | 77 | 60 | ||
| IV | 5 | 8 | 60 | .18 | 80 | .009 |
| FIGO STAGES III AND IV | ||||||
| PCI | ||||||
| PCI <10 | 13 | 27 | 85 | 38 | ||
| PCI >10 | 35 | 63 | 71 | .35 | 71 | .04 |
| Eisenkop score | ||||||
| Eisenkop <6 | 14 | 30 | 79 | 36 | ||
| Eisenkop 6-10 | 17 | 35 | 65 | 53 | ||
| Eisenkop >10 | 17 | 35 | 82 | .46 | 94 | .002 |
| Aletti score | ||||||
| Low (≤3) | 6 | 12 | 33 | 0 | ||
| Intermediate (4-7) | 15 | 32 | 67 | 53 | ||
| High (≥8) | 27 | 56 | 89 | .01 | 81 | .001 |
| Fagotti | ||||||
| Fagotti <8 | 14 | 46 | 79 | 43 | ||
| Fagotti ≥8 | 16 | 54 | 81 | .85 | 81 | .03 |
| Fagotti-modified | ||||||
| <4 | 15 | 50 | 87 | 53 | ||
| ≥4 | 15 | 50 | 73 | .36 | 73 | .26 |
| FIGO stage | ||||||
| III | 43 | 89 | 77 | 60 | ||
| IV | 5 | 11 | 60 | .41 | 80 | .39 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
