Metastatic patterns at autopsy in patients with ovarian carcinoma




Previously published studies concerning autopsy findings in ovarian cancer failed to consider the broad differences in factors that influence the course of disease. Furthermore, those studies were conducted when the currently accepted standards in diagnostics and therapy had not been fully established. The objective of the current study was to determine the frequency and sites of metastases in patients with ovarian cancer with particular attention to the clinical course and therapy.


Autopsy reports, histologic slides, and clinical files from 197 patients who died of ovarian carcinoma between 1975 and 2005 were studied. The distribution of metastatic sites (19 different organ sites) and metastatic patterns, with particular attention to clinical course (age, length of survival) and therapy (surgical treatment with curative intention, different chemotherapy regimens), were analyzed.


Overall, 66.3% of patients had metastases to sites outside the abdominopelvic cavity. Patients who were aged >70 years, who had a disease duration ≤6 months, or who had received either no treatment or treatment without curative intention more often had metastases limited to the abdominopelvic cavity. This pattern of spread was observed most frequently in patients who had received current chemotherapy regimens (odds ratio, 3.5; P = .002). Compared with patients who had received chemotherapy according to previous standards, these patients showed a significantly increased incidence of liver metastases (P < .001).


Autopsy data may yield important information concerning the metastatic potential of a malignancy and may assist physicians in making clinical management decisions. The results from the current indicated that declining autopsy rates during the last decades have limited the ability of physicians to evaluate the impact of new therapy regimens on the frequency and distribution of metastases through postmortem examination. Cancer 2007. © 2007 American Cancer Society.

In western countries, ovarian carcinoma is the fourth most frequent cause of cancer death in women and accounts for 5% of overall cancer deaths.1 In most patients, the disease is diagnosed first in its late stages. Despite current multimodal therapies, most patients cannot be cured, experience disease recurrence, and eventually die of the disease.2 According to World Health Organization estimates, there are approximately 190,000 new diagnoses of ovarian cancer annually worldwide, and approximately 114,000 patients die annually of the disease.3

Physicians involved in the care of patients with ovarian cancer, thus, often are faced with late- and end-stage disease situations. The analysis of metastatic patterns in these late stages may improve the understanding of the disease and may assist physicians in the clinical management of these patients. Only autopsy can provide an immediate and exact description of a disseminated, metastatic disease. The currently available studies of ovarian cancer at the time of autopsy have elucidated important information concerning the anatomic-morphologic extent of metastases in ovarian cancer.4–8 However, those studies hardly considered the different factors that influence the course of disease, such as age, length of survival, and treatment. Furthermore, they were performed at a time when the currently accepted standards in histopathologic diagnosis and surgical/systemic management had not yet been established fully. The current study was undertaken to determine the frequency and sites of metastasis with particular attention to these important factors.

To avoid heterogeneity, only epithelial ovarian cancers were considered in this study, because they comprise 90% of all ovarian cancers.1 The less frequent types of ovarian cancer, such as germ cell tumors and stroma cell tumors, were excluded.


Between 1975 and 2005, 22,023 autopsies were performed on women at the Institute of Pathology of the University of Basel, Switzerland. Of these, primary epithelial ovarian cancer was diagnosed in 233 women.

The autopsy reports and clinical records, including operative reports and information on chemotherapy treatment, from those 233 women were reviewed for the following information: date of diagnosis, age at death, and length of survival. To confirm the histology reported in the autopsy findings, approximately 4550 microscopic sections were reviewed, averaging 20 sections per patient.

The objective of this study was to evaluate the metastatic patterns of ovarian cancer in a homogeneous group of patients who clearly were in the final stages of disease at the time of death and to avoid the inclusion of patients who were in earlier stages of the disease and, ultimately, died of other causes. Therefore, particular attention was paid to the analysis of the clinical data and the clinical course of each patient. Thus, 37 women had to be excluded from the analysis. In 16 women, the autopsy report described ovarian cancer in an early stage; thus, ovarian carcinoma clearly was not the cause of death but, rather, an incidental finding. Nine patients had known metastatic ovarian cancer but had other, clearly identifiable causes of death based on prior clinical evaluation and confirmed by postmortem examination (accident, suicide, sepsis, cirrhosis from chronic active hepatitis, cardiac disease, and cerebrovascular disease). In 6 patients, ovarian carcinoma tissue was found in the autopsy; however, all 6 of those women died within 3 weeks after primary surgery from postoperative complications. This carcinoma tissue represented residual disease that could not be removed during debulking surgery. Six other patients were excluded because they had a second metastatic malignancy in addition to ovarian carcinoma, and their cause of death could not be attributed clearly to ovarian cancer. In addition to these 37 patients who were excluded based on clinical information, an additional 26 patients were eliminated from the current analysis after review by a specialized gynecopathologist (G.S.) showed that, in part through the use of current immunohistochemical procedures, the diagnosis of primary epithelial ovarian cancer could no longer be upheld. In these patients, metastatic disease that was identified during the autopsy could be attributed patients to other primary sources apart from the ovaries, and the site of the primary tumor could be identified retrospectively (colon, n = 10 patients; pancreas, n = 5 patients; stomach, n = 2 patients; fallopian tube, n = 2 patients; cervix uteri [squamous cell carcinoma], n = 2 patients; cervix uteri [mucinous adenocarcinoma], n = 1 patient; corpus uteri, n = 1 patient; extrahepatic bile ducts, n = 1 patient; kidney, n = 1 patient; breast, n = 1 patient).

Four of the remaining 170 patients were removed from the study group because they showed clinical characteristics that were not typical of the cohort. Their histories were extremely unusual; in particular, they had extremely long survival (300 months in 1 patient, 228 months in 1 patient, and 168 months in 2 patients). The next longest survival was 108 months.

In the end, 166 patients were available for analysis in this study. Serous carcinoma was the most common histologic type (n = 131 patients; 78.9%). The distribution of the other histologic subtypes was as follows: endometrioid carcinoma in 20 patients (12%); mucinous carcinoma in 8 patients (4.8%) and clear cell carcinoma in 7 patients (4.2%). Borderline tumors strictly were excluded. The various histologic types of epithelial ovarian cancer were not analyzed separately, because a uniform distribution of metastases and similar metastatic patterns among the different histologic types had been observed in previous studies.4, 7, 8

We recorded data regarding 19 metastatic sites. Organ sites were grouped according to whether the metastases were located in the abdominal cavity and/or retroperitoneum (I), in the thorax (II), or outside both cavities (III):

  • IMetastatic sites in the abdominal cavity and/or retroperitoneum (12 sites)
    • 1)parietal and visceral pelvic peritoneum
    • 2)parietal and visceral abdominal peritoneum, including omentum and diaphragm
    • 3)pelvic lymph nodes (LN)
    • 4)abdominal LN
    • 5)urinary bladder
    • 6)ureter
    • 7)bowel
    • 8)liver
    • 9)pancreas
    • 10)stomach
    • 11)spleen
    • 12)kidney
    For the latter 8 organ sites, prerequisites were wall invasion or parenchymal involvement; that is, involvement of the serosal surface alone was not sufficient.
  • IIMetastatic sites in the thorax (4 sites)
    • 13)LN in the thoracic region
    • 14)pleura
    • 15)lung
    • 16)pericardium
  • IIIMetastatic sites outside of both cavities (3 sites)
    • 17)bone
    • 18)brain
    • 19)skin

To evaluate metastatic patterns, the patients were organized into 7 different groups (numbers in brackets refer to the organ sites, as listed above):

  • 1Peritoneum only [1,2]
  • 2Peritoneum [1,2] and pelvic/abdominal LN [3,4]
  • 3Peritoneum [1,2] and organ sites in the abdominal cavity [5–12]
  • 4Peritoneum [1,2], organ sites in the abdominal cavity, pelvic/abdominal LN [3–12]
  • 5Abdominopelvic cavity [I] and sites in the thoracic region [II]
  • 6Abdominopelvic cavity [I] and distant sites outside thoracic cavity [III]
  • 7Abdominopelvic cavity [I], sites in the thoracic region [II], and distant sites outside both cavities [III]

Subgroups A and B: Age and Survival

To correlate the autopsy findings with clinical conditions, additional relevant parameters aside from age at death (≤70 years vs >70 years) (Subgroup A) and length of survival (≤6 months vs >6 months) (Subgroup B) were examined. Note that survival was calculated from the time of surgery or biopsy to the time of death. In patients who were diagnosed only at autopsy, the length of survival was defined as 0 months. The following therapy modes were compared.

Subgroup C: No Treatment Versus Treatment

The first group was comprised of patients who had received neither cytoreductive surgery nor systemic therapy. In many patients, the diagnosis was made first at autopsy; and, in other patients, the suspected diagnosis was made through radiologic studies or cytologic examination of ascites just a few weeks before death. In some patients, the diagnosis was confirmed by exploratory laparotomy without any extensive cytoreduction, and tissue was sampled only for the purpose of making a histologic diagnosis. These patients were compared with others who had undergone debulking surgery with curative intention. In this report, therapy with curative intention refers to the maximally achievable debulking surgery (no residual disease or residual disease in an amount that allows for the possibility of a cure through postoperative chemotherapy).

Subgroup D: Influence of Current Therapy Regimens on Distribution of Metastatic Patterns

The group of treated patients was subdivided further. We compared patients who received currently used, standard chemotherapy regimens with patients who received either no chemotherapy or cytostatic agents, such as single-agent alkylator therapy (eg, chlorambucil or melphalan), which now are used no longer in standard treatment regimens. Current systemic treatment was defined as an adjuvant therapy with ≥4 cycles of a platinum-based regimen and ≥2 additional lines of palliative chemotherapy with modern agents (eg, carboplatin, cisplatin, paclitaxel, docetaxel, liposomal doxorubicin, gemcitabine, vinorelbine, topotecan, or etoposide).

Because the number of patients in our study group was inadequate to evaluate metastatic patterns in those who had received current chemotherapies (n = 16 patients), autopsy data from 1992 to 2005 from 2 other pathology institutes (Cantonal Hospital Liestal, Switzerland and Cantonal Hospital St. Gallen, Switzerland) also were analyzed. An additional 31 patients were included who had clinical data that matched the above-described inclusion criteria. In this way, there was an adequate number of patients to establish a substudy group (n = 47 patients). The numbers of chemotherapy regimens administered to the patients in this group were 3 regimens in 27 patients, 4 regimens in 20 patients, 5 regimens in 7 patients, and 6 regimens in 3 patients.

Disease stage at first diagnosis was not considered in this analysis, because disease stage was not reported adequately according to the currently valid guidelines in many patients. However, we believe it is safe to assume that there was stage III or IV disease in the vast majority of patients, stage II disease in a few patients, and stage I disease was rare. Other studies had so few patients with early-stage disease that no meaningful statistical analysis could be performed (Dvoretsky et al. reported 9 patients with stage I disease,4 and Reed et al. reported 3 patients with stage I/II disease7). The current study was carried out in accordance with the guidelines of the Ethics Committee of the University of Basel.

Statistical Methods

Comparisons between nominal parameters were made with the Fisher exact test. The association between age at autopsy, therapy with curative intention, and current chemotherapy regimens with the metastatic pattern (ie, occurrence of metastases outside the abdominal cavity) was calculated using a multivariate logistic regression model. Statistical analyses were performed with SPSS 13.0 software (SPSS Inc., Chicago, Ill).


Clinical Characteristics of the Study Group

Table 1 summarizes clinical data from the 166 patients who were analyzed. The median age was 73 years (range, 36–98 years; mean,70 years), and the median survival of this cohort was 5 months (mean, 16 months; range, 0–198 months). Table 1 also lists clinical data for the entire cohort subdivided into the 4 subgroups based on different clinical criteria: Subgroup A, age at autopsy (>70 years vs ≤70 years); Subgroup B, disease duration (≤6 months vs >6 months); and Subgroup C, treatment options (no therapy/no therapy with curative intention vs therapy with curative intention). To assess the influence of current chemotherapy regimens on the frequency and distribution of metastases (Subgroup D), 31 patients had to be added to the main study cohort from 2 other collaborating institutions, as described above (median age, 68 years [range, 39–71 years]; median survival, 35 months [range, 12–90 months]).

Table 1. Clinical Data and Sites of Metastases of Epithelial Ovarian Cancer at Autopsy
CharacteristicNo. of patients (%)
Complete study groupSubgroup A: age at autopsy, ySubgroup B: survival, moSubgroup C: therapy with curative intentSubgroup D: chemotherapy
>70≤70≤6>6No/No therapyYesCurrent standard (substudy group)Previous standard
  • Abd indicates abdominal; Abd.pelv, abdominopelvic; LN, lymph nodes.

  • *

    The difference within the substudy group was statistically significant (P ≤ .05).

All patients1669175907684824756
Age, y
Survival, mo
Metastatic sites
 I. Abd.pelv cavity
  1. Pelvic peritoneum164 (98.8)89 (97.8)75 (100)89 (98.8)75 (98.7)83 (98.8)81 (98.8)46 (97.9)56 (100)
  2. Abd peritoneum162 (97.6)89 (97.8)74 (98.7)88 (97.7)74 (97.4)82 (97.6)80 (97.6)44 (93.6)55 (98.2)
  3. Pelvic LN46 (27.7)23 (25.3)23 (30.7)24 (26.7)22 (28.9)23 (27.4)23 (28.4)12 (25.5)18 (32.1)
  4. Abd LN123 (74.1)68 (74.7)55 (73.3)70 (77.8)53 (69.7)62 (73.8)61 (74.4)32 (68.1)40 (71.4)
  5. Urinary bladder37 (22.3)20 (22)17 (22.7)14 (15.6)*23 (30.3)*14 (16.7)23 (28.4)13 (27.7)15 (26.8)
  6. Ureter20 (12)10 (11)10 (13.3)10 (11.1)10 (13.2)8 (9.5)12 (14.6)8 (17)8 (14.3)
  7. Bowel91 (54.8)45 (49.5)46 (61.3)49 (54.4)42 (55.3)47 (55.9)44 (53.7)19 (40.4)32 (57.1)
  8. Liver80 (48.2)46 (50.5)34 (45.3)40 (44.4)40 (52.6)35 (41.7)45 (54.9)39 (83)*29 (51.8)*
  9. Pancreas12 (7.2)3 (3.3)9 (12)5 (5.6)7 (9.2)3 (3.6)9 (11)5 (10.6)7 (12.5)
  10. Stomach13 (7.8)6 (6.6)7 (9.3)3 (3.3)10 (13.2)3 (3.6)*10 (12.2)*4 (8.5)8 (14.3)
  11. Spleen27 (16.3)17 (18.7)10 (13.3)11 (12.2)16 (21.1)13 (15.5)14 (17.1)9 (19.1)8 (14.3)
  12. Kidney6 (3.6)1 (1.1)5 (6.7)3 (3.3)3 (3.9)4 (4.8)2 (2.4)4 (8.5)2 (3.6)
 II. Thoracic region
  13. LN58 (34.9)30 (33)28 (37.3)27 (30)31 (40.8)27 (32.1)31 (37.8)15 (31.9)23 (41.1)
  14. Pleura55 (33.1)23 (25.3)*32 (42.7)*25 (27.8)30 (39.5)20 (23.8)*35 (42.7)*10 (21.3)*27 (48.2)*
  15. Lung54 (32.5)27 (29.7)27 (36)23 (25.6)30 (39.5)22 (26.2)32 (39)11 (23.4)*26 (46.4)*
  16. Pericardium12 (7.2)5 (5.5)7 (9.3)4 (4.4)8 (10.5)4 (4.8)8 (9.8)0*7 (12.5)*
 III. Other
  17. Bone12 (7.2)5 (5.5)7 (9.3)6 (6.6)6 (7.9)4 (4.8)8 (9.8)2 (4.3)5 (8.9)
  18. Brain4 (2.4)3 (3.3)1 (1.3)2 (2.2)2 (2.6)2 (2.4)2 (2.4)1 (2.1)1 (1.8)
  19. Skin14 (8.4)5 (5.5)9 (12)3 (3.3)9 (11.8)4 (4.8)10 (12.2)2 (4.3)10 (17.9)

Metastatic Sites and Patterns of Spread

The frequency of metastatic disease at different sites is listed in Table 1. The peritoneal cavity was the most frequently involved site. Only 2 patients had no peritoneal involvement of disease at death: One patient had only liver involvement, and the other patient had isolated brain metastases. A comparison of the subgroups showed relatively similar distribution of metastatic sites. Compared with patients who had received chemotherapy according to previous standards, the patients who had received current chemotherapy regimens had a significantly increased incidence of liver metastases (Subgroup D: 83.0% vs 51.8%; P < .001) and less involvement of lung (23.4% vs 46.4%; P = .02) and pleura (21.3% vs 48.2%; P < .001).

Table 2 summarizes the different patterns of metastatic spread. We separated metastatic spread into 2 different patterns: metastatic burden limited to the abdominopelvic cavity and metastases that also involved the thorax and/or other distant sites. One hundred ten patients (66.3%) had metastases to sites outside of the abdominopelvic cavity. Aside from the 2 patients with isolated liver and brain metastases, all other patients with distant metastases also had intra-abdominal/peritoneal tumor involvement. Of those 108 patients, 103 not only had peritoneal involvement but also had deeper involvement of abdominal organs (wall invasion, parenchymal involvement) and/or LN.

Table 2. Distribution of Metastatic Patterns of Epithelial Ovarian Cancer at Autopsy
Characteristic*No. of patients (%)
Complete study groupSubgroup A: age at autopsy, ySubgroup B: survival, moSubgroup C: therapy with curative intentSubgroup D: chemotherapy
>70≤70≤6>6No/No therapyYesCurrent standard (substudy group)Previous standard
  • Abd indicates abdominal; LN, lymph nodes; Abd.pelv, abdominopelvic.

  • *

    Numbers in brackets refer to the organ sites as listed in Table 1.

  • The difference within the substudy group was statistically significant (P ≤ .05).

All patients1669175907684824756
Metastatic sites
 1. Peritoneum only [1,2]07 (4.2)06 (6.6)01 (1.3)4 (4.4)3 (3.9)04 (4.8)03 (3.7)04 (8.5)03 (5.3)
 2. Peritoneum [1,2], pelvic/abd LN [3,4]12 (7.2)07 (7.7)05 (6.7)9 (10)3 (3.9)07 (8.3)05 (6.1)03 (6.4)01 (1.8)
 3. Peritoneum [1,2], organ sites in the abd.pelv cavity [5–12]13 (7.8)03 (3.3)10 (13.3)5 (5.6)8 (10.5)08 (9.5)06 (7.3)06 (12.8)03 (5.3)
 4. Peritoneum [1,2], organ sites in the abd.pelv cavity [5–12], pelvic/abd LN. [3,4]24 (14.5)19 (20.9)05 (6.7)20 (22.2)4 (5.2)17 (20.2)06 (7.3)10 (21.3)02 (3.6)
Metastases limited to abd.pelv cavity [I]56 (33.7)35 (38.5)21 (28)38 (42.2)18 (23.5)36 (42.8)20 (24.4)23 (48.9)09 (16)
 5. Abd.pelv cavity [I], thoracic cavity [II]80 (48.2)42 (46.2)38 (50.7)40 (44.4)40 (52.6)37 (44)43 (52.4)18 (38.3)32 (57.1)
 6. Abd.pelv cavity [I], distant sites outside thoracic cavity [III]06 (3.6)03 (3.3)03 (4)3 (3.3)3 (3.9)03 (3.6)03 (3.7)0002 (3.6)
 7. Abd.pelv cavity [I], thoracic cavity [II], distant sites outside both cavities [III]22 (13.3)09 (9.9)13 (17.3)8 (8.9)14 (18.4)07 (8.3)15 (18.3)05 (10.6)13 (23.2)
Hepatic metastases only01 (0.6)01 (1.1)001 (1.1)0001 (1.2)000000
Brain metastases only01 (0.6)01 (1.1)00001 (1.3)0001 (1.2)01 (2.1)00

Patients who had a disease duration of up to 6 months (Subgroup B) or who had received either no treatment or treatment without curative intention (Subgroup C) had an increased incidence of metastases limited to the abdominopelvic cavity (Subgroup B: 42.2% vs 23.5%; P = .014; Subgroup C: 42.8% vs 24.2%; P = .014). This association was particularly significant when different chemotherapy modes were compared; thus, 48.9% of the patients who had received current standard chemotherapy regimens as part of their therapy had metastases only in the abdominal cavity at the time of death compared with 16% of the patients who had received therapy with previously used chemotherapy agents (P < .001).

To analyze the factors that influenced the metastatic pattern, a multivariate analysis was performed that took into account age at autopsy, therapy with curative intention, and current chemotherapy modes (Table 3). Treatment with a current chemotherapy regimen was the most significant factor for metastatic spread limited to the abdominal cavity (odds ratio [OR], 3.5; 95% confidential interval, 1.6–7.7; P = .002).

Table 3. Multivariate Analysis of Factors Influencing the Metastatic Spread Pattern Limited to the Abdominal Cavity
VariableHR (95% CI)P
  1. HR indicates hazard ratio; 95% CI, 95% confidence interval.

Current chemotherapy regimens3.5 (1.6–7.7).002
Treatment with curative intention2.3 (1.0–4.2).060
Age1.4 (0.91–2.5).151


The few studies that evaluated autopsy findings in ovarian cancer in the last 3 decades4–8 showed considerable heterogeneities in the composition of their study groups; thus, their results are difficult to compare with each other. One contributor to heterogeneity that we wanted to avoid in our study was the inclusion of nonepithelial ovarian carcinomas, such as granulosa cell tumors, germ cell tumors, and teratomas. The inclusion of these relatively rare tumors in the study group does not add significantly to the total numbers of patients analyzed, and the numbers usually are so small that these patients cannot be analyzed as an independent subgroup.

The majority of previously published studies concerning autopsy findings in ovarian cancer clearly have not correlated their findings with the clinical course of the disease. In some studies, patients were included in whom carcinoma tissue was found at autopsy, although the disease clearly was not the cause of death. Some authors included patients who did not have any extraovarian spread at autopsy.4, 6 Rose et al. even included patients who had a history of ovarian cancer, although tumor tissue was not found at autopsy. Furthermore, approximately 15% of their study group had multiple primary cancers; and, in some patients, the nonovarian cancers were related to the patients' deaths.8 We consider this practice questionable. When making statements regarding the metastatic potential of a malignancy, the clinical data also should be considered very carefully to avoid combining end-stage patients with patients in earlier stages of the disease or with patients who had only a history of ovarian cancer but had complete remission evident at autopsy. To our knowledge, our study is the first to correlate autopsy findings clearly with the clinical course of disease and to analyze exclusively patients who died directly from ovarian cancer in its final stage. Furthermore, we considered other clinical factors that significantly affected the course of the disease, namely, age at the time of autopsy, disease duration, previous treatment with curative intention, and the use of current chemotherapy regimens (factors that are not independent from each other). Most studies did not differentiate these factors and included in a single study group both young patients who had received aggressive surgical and systemic treatment and older patients with multiple comorbidities who were not considered operable or suitable for chemotherapy.4, 6, 8 Compared with those studies, Reed et al.7 were able to eliminate most of the heterogeneities in their analysis, which included 73 patients with epithelial ovarian cancer only who were treated on approved chemotherapy protocols.

In our study, we applied current knowledge of metastatic disease secondary to the ovary. The typical gross and histomorphologic features of these tumors were worked out clearly for the first time during the last decade.9–12 In particular, tumors with a strong tendency to present in the pelvic cavity may mimic, both grossly and histologically, the picture of an ovarian cancer. In these difficult and unclear tumors, misdiagnoses certainly were not uncommon. In the entire group of patients in which the autopsy diagnosis of “primary ovarian cancer” had been made (n = 233 patients), a review of the histologic slides by a gynecopathologist showed decisive morphologic features and, in selected patients, the immunohistochemical features of metastatic tumors in 11% (n = 26 patients). In these women, who were autopsied between 1975 and 1997, the primary tumor also could be identified retrospectively. The majority of studies concerning ovarian cancer in the autopsy were published during the 1970s to the 1980s.4, 6, 8 It seems reasonable to assume that those studies likewise had included approximately 10% of patients who would be diagnosed today with metastatic tumors to the ovary.

A comparison of our data with the previously published studies is difficult because of the above-mentioned heterogeneities. On the whole, our results concerning metastatic sites oscillate around the previously reported results (Table 4). Not surprisingly, our results indicate that death from ovarian cancer is associated with tumor dissemination to the peritoneum in nearly all patients. Autopsy showed distant metastases without peritoneal involvement only in 2 patients; 1 patient had exclusively liver metastases, and another patient had isolated brain metastases. Furthermore, our results indicate that the abdominopelvic cavity is not only the main site of metastatic disease but also seems to be the essential source for the development of distant metastases. In 75 of 80 patients who had metastases outside the abdominopelvic cavity, advanced tumor involvement of the intra-abdominal organs also was present, ie, tumor involvement of parenchyma (liver, pancreas, spleen, kidney), organ wall invasion (bowel, stomach, bladder, ureter), and/or involvement of LN.

Table 4. Location of Metastases in Epithelial Ovarian Cancer Found in 4 Autopsy Reviews
Period when autopsies were performedNo. of patients (%)All studies (mean)
Current study: 1975–2005Reed et al., 20007: 1972–1988Rose et al., 19898: 1958–1987Dvoretsky et al., 19884: 1958–1985
Total no. of patients166 (100)73 (100)381 (100)100 (100)720 (100)
I. Abdominopelvic cavity
 1. Peritoneum164 (99)63 (86)316 (83)073 (73)616 (86)
 2. Lymphatics123 (74)42 (58)221 (58)047 (47)433 (60)
 3. Urinary bladder037 (22)9 (12)011 (11)057 (17)
 4. Ureter020 (12)12 (16)024 (24)046 (14)
 5. Bowel091 (55)26 (36)190 (50)055 (55)362 (50)
 6. Liver080 (48)43 (59)181 (48)045 (45)349 (48)
 7. Pancreas012 (7)21 (29)044 (12)021 (21)098 (14)
 8. Stomach013 (8)077 (20)012 (12)102 (16)
 9. Spleen027 (16)20 (27)074 (19)015 (15)136 (19)
 10. Kidney006 (4)03 (4)024 (6)010 (10)044 (6)
II. Thoracic region
 11. Lymphatics058 (35)28 (39)108 (28)029 (29)223 (31)
 12. Pleura055 (33)37 (51)108 (28)038 (38)238 (33)
 13. Lung054 (33)27 (37)130 (34)039 (39)250 (35)
 14. Pericardium012 (7)12 (16)020 (5)004 (4)048 (7)
III. Other
 15. Bone012 (7)11 (15)043 (12)011 (11)077 (11)
 16. Brain004 (2)03 (4)011 (3)006 (6)024 (3)
 17. Skin014 (8)04 (5)020 (5)005 (5)043 (6)

In the analysis of patients who were treated with current chemotherapy regimens, compared with the group that received previous chemotherapy standards, we observed a significantly higher proportion of patients in which the metastatic burden was limited to the abdominal cavity. It appears that the administration of current chemotherapy agents can prevent spread of the disease outside of the abdominal cavity. The disease is more limited to the abdomen and manifests itself more frequently in the liver. This increased liver involvement may be reflective of increased survival, although it is hard to separate the direct impact of chemotherapy on the metastatic behavior of the tumor from the effect of prolonged survival that may be achieved with these agents.

In particular, our data concerning current chemotherapy regimens are difficult to compare with data from other studies. The majority of these studies accrued patients during a period when there was no effective systemic therapy for ovarian cancer or during the era of single-agent alkylating therapy. Even the results reported by Reed et al.,7 who analyzed a cohort of patients who were treated on chemotherapy protocols, are not easily comparable with our data. The autopsies that were examined by Reed et al. were performed from 1972 to 1985.7 It follows that some patients obviously had been treated on chemotherapy protocols that today are considered long out of date (eg, only 59% of patients were treated on platinum-based regimens). The approach of Reed et al. was to demonstrate the influence of therapy on metastatic patterns. However, those investigators included patients whose survival was so short (as short as 1.7 months) that there was not enough time to observe the effect of chemotherapy. Our inclusion criteria of current chemotherapy probably is better suited to represent the patients who currently are considered a “treated group of patients.”

Compared with previously reported data, Reed at al observed a higher proportion of patients with disease manifestation in liver parenchyma, pleura, and pericardium.7 We confirmed their results concerning liver involvement. However, in our study, we did not observe increased involvement of pleura or pericardium. Patients who had received chemotherapy according to a current standard had significantly less involvement of pleura and pericardium compared with patients who had received either no chemotherapy or a previously used therapy regimen. The increased number of patients who had metastatic disease in the pleura and/or pericardium after platinum-based chemotherapy led Reed et al. to the hypothesis that the tumor becomes more aggressive because of a possible mutagenic effect of the compound when therapy is not successful.7 Our data cannot support this hypothesis. In other types of cancer, it has been demonstrated that prolongation of survival because of improvements in the systemic management of the disease may result in an increased number of patients who develop distant metastases, as demonstrated by the increased rates of brain metastases in patients with colon cancer and patients with breast cancer who were treated on current therapy regimens.13, 14 This does not appear to be true for patients with ovarian cancer.

The metastatic patterns noted by Reed et al., which clearly differ from our findings, may be explained in part by their relatively young cohort (only 3 of their patients were aged >70 years at diagnosis).7 However, on average, patients who die of ovarian cancer are older. In our study cohort, we had a high number of older patients who were diagnosed first with ovarian cancer at autopsy. This explains why the patients in our entire study group had a relatively short median survival of 5 months compared with that reported by Rose et al. (14 months)8 and by Reed et al. (15 months).7 These patients comprised the majority of our study Subgroup C (survival ≤6 months). Thus, analysis of the metastatic patterns observed in this group is noteworthy, because these patients demonstrate the disease in its natural course (ie, not affected by any therapy).

It has been discussed widely that autopsy rates have decreased dramatically in many countries.15–18 This also was reflected in our study cohort. From the first decade of the study period (1975–1984), 91 patients were recruited; whereas, from 2002 to 2005, only 5 patients were recruited (the 10-year periods were 1975–1984, n = 91 patients; 1985–1994, n = 49 patients; and 1995–2004, n = 26 patients). Because of the reduced number of autopsies, important information regarding the morphologic extent of disease probably will be missed.18–21 In particular, the development of metastatic patterns under new therapy regimens probably will be detected only through clinical and radiologic investigations in the future. A postmortem confirmation through direct gross and histologic observation will occur only in rare individual patients (a typical result of the current autopsy culture is our need to obtain data from 2 additional pathology institutes to have a sufficient number of patients to evaluate the impact of current chemotherapy regimens). It is questionable whether clinical studies alone can comprehensively answer such questions. Compared with studies in which the frequency and extent of distant metastases in patients with ovarian cancer were evaluated through autopsies, studies that used only clinical data from living patients22, 23 showed a consistently lower incidence of metastatically involved organs. With the current study, we wanted to underline the clinical and scientific importance of autopsy findings, and we hope that it will contribute to the resumption of the postmortem examination as an important tool for the evaluation of oncologic diseases.


The authors thank Gieri Cathomas, MD (Institute of Pathology, Cantonal Hospital Liestal, Switzerland) and Pierre Andre Diener, MD (Institute of Pathology, Cantonal Hospital St. Gallen, Switzerland) for providing the autopsy reports from their departments and Andreas Schotzau, MS (JPS Institute for Biomathematics, Basel, Switzerland) for statistical support