Ovarian metastases from breast cancer

Report of 29 cases

Authors


Abstract

BACKGROUND:

The objective of this study was to describe the characteristics and survival outcomes of patients with breast cancer who had ovarian metastases.

METHODS:

Data from 29 women who underwent surgery were reviewed retrospectively (from 1998 to 2007). Patient characteristics, tumor characteristics, and treatment data were collected. Pelvic extent of disease was documented using a system analogous to the International Federation of Gynecology and Obstetrics classification for ovarian cancer. Global survival, disease-free intervals, and the distribution to other metastatic sites over time were studied. Outcomes were compared between the group who underwent macroscopic resection of lesions and the group who did not undergo resection.

RESULTS:

The data indicated a predominance of premenopausal and hormone receptor-positive status and a greater prevalence of lobular infiltrating carcinoma, bilateral breast cancer, and predisposing genetic factors compared with the global population with breast cancer. Ovarian disease was diagnosed at a median of 5 years after breast cancer. Seventy-five percent of patients were asymptomatic, and advanced-stage pelvic extent or extra-abdominal metastases were observed in 41.5% of patients. The median survival was 3 years, and the median follow-up was 2 years. Survival improved significantly when optimal debulking surgery was performed.

CONCLUSIONS:

Breast cancers may be associated with ovarian metastases. The current results indicated that surgical resection tends to increase survival, which may be long; however, larger series would be needed to confirm other prognostic factors. The high rates of hormone receptor-positive tumors and premenopausal patients led the authors to suggest that the surgical option should consist of at least bilateral oophorectomy, even when the contralateral ovary appears to be normal. Cancer 2010. © 2010 American Cancer Society.

Less than 10% of patients with breast cancer (BC) have evidence of distant metastases at diagnosis, but 30% of them will have recurrent, metastatic disease.1 Follow-up after BC and screening for distant metastases should become the focus of increased concern because of a predictable improvement in BC survivorship in developed countries.2, 3 Metastases and micrometastases in the ovaries have been reported with a prevalence ranging from 3% to 30% in various series, including autopsies, prophylactic or therapeutic oophorectomies, and incidental findings in routine surgery.4-15 The specifics of breast primary tumors with ovarian metastases (OM) are unknown, although a greater proportion of invasive lobular carcinoma was reported.10

In clinical practice, an adnexal mass detected in a patient who has a previous history of BC reportedly is related more often to benign disease. When such a mass is malignant, it originates from a primitive ovarian carcinoma 3 to 7 times as often as it originates from metastatic disease.11, 16, 17 In such patients, surgical biopsies are required to establish a differential diagnosis between a primitive ovarian cancer and metastatic disease. Further choices may include therapeutic ovarian ablation or surgical resection of metastatic pelvic disease. If the overall prognosis appears to be similar to that in patients with other visceral metastases, with a median survival of 2 years, then survival for >10 years has been reported.4, 18 Recent studies have indicated a possible survival benefit after cytoreductive surgery; however, the criteria for patient selection remain unclear.7-9 Are metastatic disease stage and disease-free interval between primary breast tumor and metastatic disease prognostic factors that could help in the decision to undergo surgery? The objective of this retrospective study was to describe the characteristics of patients with BC patients who had ovarian metastatic disease.

MATERIALS AND METHODS

Data from 5 institutions concerning 29 women who had documented OM from breast carcinoma between 1998 and 2007 were reviewed retrospectively. Information was sought for each patient on age at BC diagnosis, menopausal status, personal and familial history of cancer, and BC gene (BRCA) mutation screening. Available data on primary BC, pelvic extent, circumstances of diagnosis, and therapeutics were collected. To characterize BC, variables included the histologic type of BC; if patients had bilateral BC, then whether or not it was synchronous; staging (pathologic tumor, lymph node, metastasis [pTNM] classification); grading (Scarff-Bloom-Richardson [SBR] grade); hormone receptor and HER2/neu status; and whether or not patients developed local recurrence. For metastatic extent, variables included the number and location of metastases, time intervals to the detection of pelvic and extrapelvic metastases, hormone and HER2/neu receptor status when available, and staging. For each patient, the extent of metastases in the pelvis was expressed using the operative and pathologic findings analogous to the International Federation of Gynecology and Obstetrics (FIGO) staging for ovarian carcinoma.

On the basis of formerly published studies, surgical options were divided in 2 groups. The first group included patients who had nonoptimal surgical options from biopsies to partial resections, and the second group included patients who underwent macroscopic optimal resection with or without lymphadenectomy. Global survival was studied for all patients and for each group separately using Kaplan-Meier curves. Both groups were compared for survival and for variables that could influence survival (disease-free interval >5 years, pelvic extent, and the presence of other metastatic sites). We used the log-rank test (P < .05) to compare survival outcomes and univariate analysis with chi-square or Fisher exact tests, when appropriate, to compare variables between groups.

RESULTS

At the time of BC diagnosis, the median age of the 29 patients was 48 years (range, 33-67 years), and 11 of 29 patients (62%) were premenopausal. Seven of 29 patients (24%) had a bilateral BC, which was synchronous in 3 of 29 patients (10%). BRCA mutation screening data were available and positive in 3 of 29 patients (10%; 2 patients had BRCA2 mutations, and 1 patient had a BRCA1 mutation). All patients who had a BRCA gene mutation identified had a bilateral BC. Another patient had a synchronous bilateral BC and a family history of BC, but BCRA screening was not available in our data. Histologic types were ductular invasive carcinoma in 16 of 29 patients (55%), lobular invasive carcinoma in 12 of 29 patients (43.5%), and not known for 1 patient. The majority of patients had a locally advanced disease stage and positive axillary lymph nodes, and in 20 of 29 patients (69%) had tumors that measured >2 cm. However, only 1 of 29 patients had stage IV disease with bone metastases. Grade was SBR 2 and 3 in 18 of 29 tumors (62%) and 3 of 29 tumors (10%), respectively. Estrogen receptor status was positive in 23 of 29 tumors (79.5%). HER2/neu expression was detected in 3 of 12 tested samples.

All patients underwent surgical resection of the initial breast tumor with axillary dissection by either mastectomy (20 of 29 patients) or lumpectomy (9 of 29 patients). The majority of patients received adjuvant therapy: Twenty-eight of 29 patients (96.5%) received radiation, 22 of 29 patients (76%) received adjuvant chemotherapy, and 10 of 29 patients (34%) received neoadjuvant chemotherapy. Hormone therapy was received by 9 of 11 postmenopausal women (82%) and by 8 of 18 premenopausal women (44%), and castration was added for 5 of those 8 women (28%). A minority of patients (2 of 29 women) were given trastuzumab. Eight of 29 patients (27.5%) had developed a local recurrence.

The median time between primary BC diagnosis and the diagnosis of pelvic disease was 5 years (range, 0-20 years). There were no guidelines for a routine follow-up after BC except for clinical examination. However carbohydrate antigen 15.3 (CA 15.3) levels were checked regularly in almost all patients. In all patients except 1 who had an incidental finding of ovarian lesions during colectomy for a villous polyp, pelvic disease was suspected on radiologic findings that led to surgery. In 75%, patients were asymptomatic, and a routine radiologic checkup was performed for an elevation of tumor markers. Nine of 29 patients (31%) underwent unilateral oophorectomy or ovarian biopsy with or without peritoneal biopsies; 10 of 29 patients (34.5%) underwent bilateral oophorectomy and peritoneal biopsies with or without omentectomy; and 10 of 29 patients (34.5%) underwent multiple resections added to oophorectomy, including hysterectomy in 8 of 29 patients, with or without bowel resection or appendectomy in 5 of 29 patients, and with pelvic and para-aortic lymphadenectomy in 1 patient.

Eighteen of 29 patients (62%) underwent nonoptimal surgical resection, whereas 11 of 29 patients (38%) patients underwent optimal macroscopic resection according to the operative reports. Pathologic findings indicated bilateral ovarian disease in 22 of 29 patients (76%), unilateral disease in 4 of 29 patients (14%), and no sufficient data to assess unilaterality in 3 of 29 patients (10%; biopsies or former oophorectomy). Two of 8 patients (25%) who underwent hysterectomy had uterine micrometastatic involvement. When surgery for pelvic disease was performed, 12 of 29 patients (41.5%) had other metastatic sites identified (if it was believed that these patients had a primitive ovarian cancer, then it would be staged as FIGO stage IV ovarian carcinoma). The same number of patients had local disease extent that was not limited to pelvis (if it was believed that they had a primitive ovarian cancer, then it would be staged as FIGO stage III ovarian carcinoma). Only 5 of 19 patients (17%) had disease extent that was categorized as limited to the pelvis (if it was believed that they had a primitive ovarian cancer, then it would be staged as FIGO stage IIA or IIB ovarian carcinoma). In these early stages, CA 15.3 levels were normal at the time of diagnosis in 3 of 5 patients, 4 of 5 patients had benign cysts on radiology, and 4 of 5 patients had lobular carcinoma. Hormone receptor status in metastases was available in 12 patients and was positive in 11 of those 12 patients (92%). Twenty-seven of 29 patients (93%) received salvage chemotherapy, which contained docetaxel in 55% of patients, and more than 1 line was received by 69% of patients. Twelve of 29 patients (41.5%) received hormone therapy. Two patients received external radiation.

By 1 year after the assessment of pelvic metastases, extrapelvic metastases were diagnosed in 11 of 17 patients (65%) who initially were free of extrapelvic metastases (liver metastases in 50%). At a median follow-up of 2 years, the median global survival was 3 years (range, 0.5- 9 years) (Fig. 1). Statistical significance was reached in a comparison between patients who underwent nonoptimal surgical resection (median survival, 2 years) and patients who underwent optimal resection (median survival, not reached; P = .015; log-rank test) (Fig. 2). A metastatic extent limited to the pelvis (analogous to FIGO stage II ovarian carcinoma) was observed significantly more often in the resection group than in the nonresection group. In univariate analysis, no statistical significance was reached for other studied variables that had the potential to influence survival between the 2 groups (disease-free interval >5 years, classification analogous to FIGO stage III or IV, and a diagnosis of other metastatic sites in the following year), and there was no difference between these groups in histologic type (Table 1). Survival did not differ significantly with a time to recurrence of <5 years or >5 years.

Figure 1.

This chart illustrates global survival (median follow-up, 2 years). CI indicates confidence interval.

Figure 2.

This chart illustrates survival according to surgical resection (P = .015). The blue line represents the resection group in which 12 of 29 patients (41.5%) had other metastatic sites (if it was believed that these patients had a primitive ovarian cancer, then it would be classified as International Federation of Gynecology and Obstetrics [FIGO] stage IV ovarian carcinoma). The same number of patients had local disease that was not limited to pelvis (if it was believed that these patients had a primitive ovarian cancer, then it would be classified as FIGO stage III ovarian carcinoma). Only 5 of 19 patients (17%) had disease that was limited to the pelvis (if it was believed that these patients had a primitive ovarian cancer, then it would be classified as FIGO stage IIA and IIB ovarian carcinoma). The red line represents the resection group, in which patients underwent optimal macroscopic resection of metastases.

Table 1. Comparison of Pelvic Extent Characteristics and Histologic Type of Breast Cancer in the Resected Group and the Nonresected Group: Univariate Analysis (N = 29)
CharacteristicNo. of PatientsP
No Resection, n = 18aResection, n = 11b
  • FIGO indicates International Federation of Gynecology and Obstetrics.

  • a

    Patients who underwent nonoptimal surgical resection of metastases.

  • b

    Patients who underwent optimal macroscopic resection of metastases.

Ovarian metastases classified as a FIGO stage II primitive ovarian cancer05.004
Ovarian metastases classified as a FIGO stage III primitive ovarian cancer102.27
Ovarian metastases classified as a FIGO stage IV primitive ovarian cancer84.9
Extrapelvic metastatic site by 1 y after diagnostic of pelvic metastases53.65
Disease-free interval between diagnosis of breast cancer and ovarian metastases >5 y96.56
Lobular invasive carcinoma66.13

DISCUSSION

We report the third largest series of patients who underwent surgery for OM from BC, including 29 cases over the shortest period (9 years). Eitan et al.7 and Ayhan et al.8 reported, respectively, 59 patients aged >16 years and 35 patients aged >22 years. In the literature, prevalence varied over a wide range, from 1% in the series by Curtin et al.11 (incidental adnexal surgery) to 13% (surgery for pelvic mass) and up to 25% (autopsies or therapeutic oophorectomy),9, 11-13, 15, 16 reflecting different stages of the disease's course. Cases were spread over long periods (from 6 to 36 years), so their incidence may be rare. The proportion of micrometastases reached 46% in the series reported by Gagnon and Tetu.5 More recently, ovarian micrometastases were reported in 3.4% of systematic biopsies in a surgical sparing fertility program before patients received chemotherapy for BC.

Primary BC in an advanced stage (69% of patients had at least pT2N1 disease) and a median age at BC diagnosis of ≈50 years in our series were similar in the studies by Eitan et al.,7 Ayhan et al.,8 and Le Thai et al.14 A greater proportion of lobular subtype was observed in our patients than in the global population with BC (43% vs 4%-15%) and in the series reported by Le Thai et al.14 and Eitan et al.7 (12% and 22%, respectively). However, a high rate of lobular carcinoma (63%) was reported in patients who had BC with endocrine metastases detected at autopsy.16 No systematic review of hormone receptor status in patients who had primary BC with OM was provided in former studies. In the general BC population, hormone receptors are positive in 70% of patients. High rates of hormone receptor-positive BC (79.5%) and of premenopausal status (62%) in our data may suggest the presence of hormone regulation in the development of OM. Lee and Hori15 raised this idea when they observed OM more often in functional ovaries. Our data indicated a prevalence of BRCA mutations significantly superior to that in the general population of women aged <50 years with BC (range, 1.2%-1.4%).19 Tserkezoglou et al.20 compared the statistical risk of mutation (using a BRCAPRO computer model) in primitive and metastatic ovarian cancer after BC and reported results similar to ours for the latter (41% and 9%, respectively). The paradox of ubiquitous expression of BRCA proteins and selective risk to the breasts and ovaries may be explained by an interaction with estrogens.21 In addition, our data revealed a greater proportion of bilateral BC (24%, 10% of which were synchronous).

Seventy-five percent of our patients were asymptomatic and had a radiologic checkup either as a matter of routine or because of elevated tumor markers. Eitan et al.7 reported that 28 of 59 patients (47%) were without any symptoms at the time of OM diagnosis, 15 of 59 patients (25%) had gastrointestinal symptoms, but only 11 of 59 patients (19%) had abdominal pain. Ayhan et al.9 reported that 3 of 35 patients (9%) had no symptoms, 19 of 35 patients (54%) had abdominal distension, 3 of 35 patients (9%) had pressure symptoms, 6 of 35 patients (17%) had an abdominal pelvic mass, 3 of 35 patients (9%) had abnormal uterine bleeding, and only 1 of 35 patients (3%) had abdominal pain. Those authors did not indicate precisely what initially lead to the diagnoses of OM (ie, the symptoms or the checkups). Furthermore, in those studies, a small proportion of patients had abdominal pain; the main symptoms were gastrointestinal in the study by Ayhan et al. and abdominal distension in the study by Eitan et al. and may have been paucisymptomatic. We believe that it is important to emphasize that the absence of symptoms or the presence of minor symptoms may explain a diagnosis of advanced, metastatic pelvic disease stage a long time after the initial BC. Asymptomatic forms, a long disease-free interval between breast and ovarian disease assessment, and advanced stage pelvic disease in our data and in former studies7, 8, 16 raised the issue of the screening for OM as early as possible. Although elevation of CA 15.3 may precede the assessment of distant metastases in 2 of 3 of patients,22 in our data, CA 15.3 levels were normal in 3 of 5 patients with early stage pelvic disease (II). Cysts >5 cm on ultrasound (US) reportedly were associated more with primitive ovarian cancer than with OM, in which ovaries could appear either normal or as solid tumors.17 In our data, ovarian cysts in early pelvic disease stage (II) for 4 of 5 patients detected on US led to surgery, but the slides revealed benign cysts coexisting with OM. The detection of pelvic metastases with [18F]fluorodeoxyglucose-positron emission tomography tomodensitometry was feasible in 4 women who were in clinical remission after BC23 without any correlation with other radiologic examinations. In a case report, Fondrinier et al.24 emphasized the possibility of the misdiagnosis of an initial metastatic BC with peritoneal involvement in lobular cancer in which a routine checkup was negative, and they discussed the role of laparoscopy for the staging in the lobular type. In our data, 4 of 5 patients with early stage pelvic disease had had a primary lobular invasive type. Other than our current study, 3 other retrospective studies7-9 reported a trend toward a significant improvement in survival when patients underwent optimal surgical resection of pelvic metastases. On multivariate analysis, a time to recurrence >5 years and optimal debulking were identified as significant factors.7 In our data, the time to recurrence did not differ significantly for the group with longer survival. Although Eitan et al.7 noted that the presence of extrapelvic disease decreased significantly the chances of achieving optimal debulking, none of the precedent studies specified survival in relation to pelvic stages. In our data, all of the patients with stage II disease (n = 5) were in the group who underwent optimal resection and had significantly longer survival, but there was no significant difference between the 2 groups concerning the distribution of stages III and IV. Eitan et al.7 reported a significant improvement in survival for patients who underwent optimal surgical resection but reported no significant improvement in survival when they compared patients with the only criteria of intra-abdominal disease outside the pelvis versus no intra-abdominal disease outside the pelvis. In the absence of systematic lymphadenectomy, using a classification analogous to the FIGO staging classification based on operative findings could result in an underestimation of the metastatic extent. In fact, lymph nodes were positive in 69% of patients in the series by Ayhan et al.,8 in which 100% of patients underwent lymphadenectomy. In the literature, the longest survival reported after OM with clinical remission was 12 years in a the patient who underwent radical lymphadenectomy and for stage I disease (unilateral OM).18 Metastases outside the pelvis at the time of surgery were present in 41.5% of patients in our data, in 44% of patients in the study by Eitan et al.,7 and in 6% of patients in the study by Ayhan et al.8. To our knowledge, the influence of apparition of other metastases after pelvic surgery on survival has never been studied. In our data, 65% of the patients who had no extrapelvic metastases at the time of surgery developed extrapelvic metastases in the following year.

In conclusion, surgical resection tends to increase survival, even in patients who have an advanced pelvic disease stage; however, the influence of other prognostic factors on survival remains unclear. Larger series will be needed to conduct a multivariate analysis of prognostic factors. To our knowledge, surgical resection of pelvic metastases after chemotherapy has never been evaluated. High rates of hormone receptor-positive tumors and premenopausal patients in our data led us to suggest that surgery should consist of at least bilateral oophorectomy, even if the contralateral ovary appears to be normal.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary