Ascites as a predictor of ovarian malignancy.

OBJECTIVE

To determine the utility of ascites as a predictor of ovarian malignancy and define its relationship with the histologic type of ovarian tumor (benign, borderline, or malignant) and stage of disease.

METHODS

This retrospective cohort study analyzed the clinical and pathological finding of 125 patients from two institutions treated for a pelvic mass. Preoperative data to include: physical examination, imaging studies (USD, CT, or MRI), and operative reports were reviewed for evidence of ascites. This was correlated with final pathologic findings and stage of disease. Collected data were summarized with descriptive statistics. Further statistical analysis was performed using Pearson's chi(2), cross tabulation, and the Median Test. Data were analyzed with SPSS 6.1 for Windows.

RESULTS

One-hundred twenty-five patients were evaluable for this study. The ovarian pathologic findings were as follows: 57 benign (45%), 12 borderline (10%), and 56 malignant (45%). Fifty-three patients (42%) had frank ascites at laparotomy. Seventy-two patients (58%) had no ascites. All patients with ascites diagnosed preoperatively (n = 41) on physical examination or imaging studies were confirmed intraoperatively. Absence of ascites was correctly diagnosed preoperatively in 72/84 patients (86%). Of the 57 benign tumors, only 5 patients (9%) had small amounts of peritoneal effusion. Of the 12 borderline tumors, 7 patients (58%) had ascites. Of the 56 malignant tumors, 41 (73%) had ascites. Using presence or absence of ascites on clinical assessment as the predictor variable and benign or malignant (borderline and invasive histopathology) tumors as the outcome variable, the positive predictive value (PPV) of ascites to detect ovarian malignancy was 95% and the negative predictive value (NPV) was 64%. When borderline tumors were excluded, the PPV and NPV of ascites to detect malignant invasive tumors were 95 and 73%, respectively. Furthermore, a progressive relationship between stage of ovarian malignancy and percentage of cases with ascites was identified. Ovarian malignancies in the early stages (I and II) produced ascites only in 17% of the cases. In advanced stages (III and IV), 89% produced ascites. In addition, for stage I and II disease, all patients possessed <0.5 liters of ascites at surgery, whereas the majority of patients (66%) with stage III and IV disease had >0.5 liters.

CONCLUSIONS

Our findings indicate the presence of ascites on preoperative physical examination or imaging study is highly predictive of ovarian malignancy in women with a pelvic mass. The absence of ascites may not always predict benign disease since nearly half of borderline tumors and 83% of early stage malignant ovarian tumors do not produce ascites. A progressive relationship between stage of malignancy and incidence as well as volume of ascites was also observed.