Aim To compare distinct clinicopathological features between atypical proliferative serous tumors and non-invasive low-grade ovarian serous carcinomas.
Methods Our study group comprised 203 cases of serous borderline tumors sub-classified as atypical proliferative serous tumors or as non-invasive low-grade serous carcinomas. All pathological features related to borderline tumors were re-evaluated by two gynecological pathologists. Data concerning recurrences and survival were retrieved from the medical records of the patients.
Results When comparing atypical proliferative serous tumors to non-invasive low-grade serous carcinomas, the latter were statistically related to advanced stage at diagnosis, bilateral disease, exophytic pattern of growth, microinvasive carcinoma, and the presence of invasive implants. In univariate analysis, recurrences were statistically related to the exophytic pattern of growth, to microinvasion, and to the presence of implants (both invasive and non-invasive). Nevertheless, in multivariate analysis, only microinvasion and the presence of invasive implants were related to recurrence. Women who eventually succumbed to the disease were only those with invasive implants. Their ovarian tumor was either a non-invasive low-grade serous carcinoma or an atypical proliferative serous tumor with ‘minimal’ micropapillary pattern. Neither lymph node involvement nor endosalpingiosis seemed to influence the course of the disease.
Conclusions The results of our study underline the increased possibility of non-invasive low-grade serous carcinomas to be related with features indicative of aggressive behavior as opposed to atypical proliferative serous tumors. Nevertheless, irrespective of tumor histology, the presence of invasive implants and microinvasion were the only independent prognostications of recurrence.
- serous borderline tumors
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In multivariate analysis, only microinvasion and invasive implants were found to be a risk factor for recurrence.
Women who succumbed to the disease were only those with invasive implants, irrespective of tumor sub-type.
The indolent nature of ‘minimal’ micropapillary areas found in atypical proliferative serous tumors is questioned.
Since their first description by Taylor in 1929 as semi-malignant,1 several terms have been used to classify the group of serous tumors displaying an intermediate behavior between a benign cystadenoma and a malignant serous carcinoma. Through the years these tumors have been designated as tumors of borderline malignancy/carcinomas of low malignant potential2 or as atypical proliferative tumors,3 with the last World Health Organization (WHO) 2014 classification4 introducing the term of ‘borderline serous tumor’ of usual or micropapillary variant—those terms being considered synonymous with atypical proliferative serous tumor and non-invasive low-grade serous carcinoma, respectively.
In the recent literature, several studies have attempted to identify specific clinicopathological features of serous borderline tumors, which could be associated with an adverse prognosis.5–7 The importance of stratifying these tumors into low- and high-risk for recurrence is strongly related to the fact that approximately one third of these tumors are found in young women wishing to preserve their fertility. Yet fertility preservation, along with the stage of the disease, the incomplete surgical staging, and the existence of residual tumor, has already been confirmed to be an independent prognostic factor for disease recurrence.8 Therefore, multiple studies are needed that would incorporate recently acquired knowledge, through the changes implemented by the new WHO 2014 classification, with emphasis on areas with ongoing controversy.9
The purpose of the present study was to evaluate the clinical behavior of serous borderline tumors both of usual and micropapillary subtype, in a large cohort of patients, and relate it to several clinicopathological parameters. Changes in evaluating features such as microinvasion versus microinvasive carcinoma, recently incorporated into the last WHO classification,4 or the importance of lymph node involvement and endosalpingiosis, will be considered. Finally, taking into consideration the results of the previously mentioned study, we aim to highlight our supposition on the clinical significance of minimal micropapillary areas in an otherwise usual serous borderline tumor/atypical proliferative serous tumor.
Two hundred and seventy-two consecutive unselected cases of serous borderline tumors operated between January 2001 and September 2017 were retrieved from the files of our pathology department. The respective slides were reviewed by two pathologists specialized in gynecological pathology (IM and DC) who were blinded to the clinical outcome of the patients. Following the protocol used in the department for ovarian borderline tumors, one section per centimeter of maximal tumor diameter for tumors smaller than 10 cm was obtained, and two sections per centimeter for larger tumors. All pathological features were re-evaluated following the WHO 2014 criteria.9
Information concerning the clinical follow-up of the patients was obtained from the records of the hospital or after personal communication. Forty cases were lost to follow-up while 29 were considered as serous cystadenomas with focal epithelial atypia/epithelial hyperplasia10 and were not included in the analysis. The remaining 203 cases were sub-classified as atypical proliferative serous tumors or as non-invasive low-grade serous carcinomas. In 143 patients, a surgical staging was performed, consisting of omentectomy followed by multiple peritoneal biopsies, while in 52 of them a lymphadenectomy was also conducted. Patients with invasive implants/low grade peritoneal serous adenocarcinoma received adjuvant chemotherapy consisting of paclitaxel followed by carboplatin.
The age of the patients, the uni- or bilateral localization of the tumor, the exo- or endophytic pattern of growth, the presence of microinvasion or microinvasive carcinoma, and the presence of peritoneal implants, lymph node involvement, or concurrent endosalpingiosis were recorded.
All evaluated clinicopathological characteristics are presented in Table 1.
Written informed consent was obtained from all patients and the study was approved by the local ethics committee.
Definitions and details of the pathologic criteria were as follows. Micropapillary pattern/non-invasive low-grade serous carcinoma was characterized by the presence of micropapillary or cribriform features with scant or no stromal cores, measuring at least 5 mm in one dimension. The cells should be relatively uniform with a higher nuclear atypia or mitotic index as compared with the cellular population of atypical proliferative serous tumors.11 Atypical proliferative serous tumors were considered as with ‘minimal’ micropapillary pattern, when characterized by the identification of small areas of micropapillary growth, each one not measuring >5 mm in its largest diameter.12 Microinvasion was only used for isolated rounded eosinophilic cells or cell clusters within the stroma that measured <5 mm in greatest diameter. They were often surrounded by retraction spaces and resembled the epithelial cells lining the papillae. Cases with higher cytologic atypia and a cribriform or solid growth, even of small size (<5 mm), were reclassified as ‘microinvasive carcinomas’.9 Implants, formerly designated as ‘non-invasive’, were reclassified as ‘implants’, and any ‘invasive’, as characterized by the infiltration of the underlying stroma, were characterized as low-grade serous carcinoma.13 For tissue invasion, the expanded criteria as proposed by Bell et al14 were also used. Specifically, implants with a micropapillary or cribriform pattern or solid nests surrounded by a clear space were also considered as invasive. Finally, lymph node involvement was considered when individual cells with cytoplasmic eosinophilia, or clusters of cells or non-branching papillae, were found within the lymph node sinuses or the parenchyma, without destructive invasion.15
Photomicrographs illustrating the diagnostic terms are provided in Figures 1 and 2.
Statistical Package for the Social Sciences (SPSS) version 24.0 was used to perform all statistical analyses. Pearson χ2 test and Fisher's exact tests were used to compare statistically significant differences between categorical variables (tumor type, size, tumor laterality, pattern of growth, microinvasion, microinvasive carcinoma, implants, lymph node involvement, and endosalpingiosis). The t-test was applied to compare continuous variables (age). The Kaplan-Meier method was used for recurrence curves. Cox regression analysis (univariate and multivariate) was used to assess the effects of different variables on recurrence rates. Any statistical relationship of p<0.05 was considered statistically significant.
Of the 203 patients who were found eligible for the analysis, 163 had an atypical proliferative serous tumor (80%), while in 40 (20%) the tumor had the features of a non-invasive low-grade serous carcinoma. Median follow-up was 110 months (range 12–212).
All statistical results are presented in Table 2.
The median age at diagnosis was 42 years (range 19–76) in women with atypical proliferative serous tumors and 44 years (range 21–76) in women with non-invasive low-grade serous carcinomas (p=0.2).
The median tumor size was 5.8 cm (range 0.5–23) in women with atypical proliferative serous tumors and 5.6 cm in women with non-invasive low-grade serous carcinomas (range 0.7–28) (p=0.6).
Bilateral disease was significantly more common in women with non-invasive low-grade serous carcinomas (60%, n=24) than in women with atypical proliferative serous tumors (28.2%, n=46) (p<0.001).
The exophytic pattern of growth was recognized approximately two times more often in non-invasive low-grade serous carcinomas than in atypical proliferative serous tumors (42.5% and 24.5%, respectively, p=0.024).
Microinvasion was observed in 21 cases (10%) while seven cases (3.4%) were reclassified as ‘microinvasive carcinomas’. Microinvasion was encountered in 16 atypical proliferative serous tumors (9.8%) and in five non-invasive low-grade serous carcinomas (12.5%) and this difference was not of statistical significance.
Higher rates of microinvasive carcinoma were found in non-invasive low-grade serous carcinomas (15%, n=6/40) as opposed to atypical proliferative serous tumors (0.6%, n=1/163) (p<0.001).
As stated above, only 143 women were adequately staged, either in the initial operation or after their first recurrence. Of these, 113 had an atypical proliferative serous tumor and 30 had a non-invasive low-grade serous carcinoma. In 52 of them, pelvic ± para-aortic lymphadenectomy was also performed. The median number of excised lymph nodes was 14 (range 1–46).
Stage I disease was encountered in 87 (77%) atypical proliferative serous tumors and in 15 (50%) non-invasive low-grade serous carcinomas. Twenty-six (23%) atypical proliferative serous tumors and 15 (50%) non-invasive low-grade serous carcinomas were of higher stage (II–IV) (p=0.006, Fisher’s exact test).
Peritoneal implants were identified in 41cases (29%), of which nine were invasive (6%) and 32 non-invasive (23%). Invasive implants were diagnosed significantly more frequently in women with non-invasive low-grade serous carcinoma (20%) than in women with atypical proliferative serous tumor (2.7%) (Figures 3 and 4).
Lymphadenectomy was performed in 52 women and there were seven women diagnosed with lymph node involvement. In all cases lymph node disease was in the form of sinus eosinophilic cells or papillary tufts resembling an atypical proliferative serous tumor (Figure 2). No metastatic low-grade serous carcinoma was encountered. No statistically significant association was found between lymph node involvement, tumor subtype, the presence of microinvasion or microinvasive carcinoma, and the pattern of tumor growth (p=0.4, p=0.68, p=1, and p=0.45, respectively, Fisher’s exact test). Lymph node involvement was strongly associated with the presence of peritoneal implants (p=0.003, Fisher's exact test).
Endosalpingiosis was reported in 47 of the 143 surgically staged cases. It was not statistically related to the type of tumor (p=0.664) or to the identification of implants, either invasive or non-invasive (p=0.210).
During our study period, there were eleven recurrences. Four of them had the features of invasive implants. In univariate analysis recurrences were statistically related to the exophytic pattern of growth (p=0.015), to microinvasion (p=0.0004), and to the presence of peritoneal implants, both invasive and non-invasive (p=0.009). Neither the type of tumor nor lymph node involvement or endosalpingiosis were significantly associated with recurrence. In multivariate analysis only microinvasion (p=0.0002) and invasive implants (p=0.0004) were independent negative prognosticators for recurrence (Table 3). Kaplan-Meier curves are presented in Figure 4.
One woman with recurrent disease is still alive, 112 months after the initial presentation and 9 months after her third recurrence which had the features of a low-grade serous carcinoma. Also, three patients with invasive implants/low-grade serous carcinoma either in the first operation or in the recurrence died of disease 44.69 and 86 months after the initial presentation. Their ovarian tumor had either the features of an atypical proliferative serous tumor with minimal areas of micropapillary growth (two cases) or of a non-invasive low-grade serous carcinoma (one case). Moreover, all these three tumors had an exophytic pattern of growth. Due to the limited number of deaths, analysis for overall survival could not be performed.
At the time of last contact, five patients had died from unrelated causes, while all the remaining patients were alive and disease-free.
For many years, the terminology, biological behavior and, consequently, surgical approach for ovarian serous borderline tumors has been the subject of research for many investigators.10 Recently, two population-based studies, that analyzed data from the Danish Pathology Data Bank or the Danish Cancer Registry, have highlighted issues such as the risk for advanced stage disease and invasive implants or the overall survival, depending on the specific sub-type of serous borderline tumor—either an atypical proliferative serous tumor or a non-invasive low-grade serous carcinoma.11 16 Nevertheless, in the above studies, features such as lymph node involvement, the possible predictive role of endosalpingiosis in tumor recurrence, or the importance of identifying ‘minimal’ micropapillary areas in the setting of an atypical proliferative serous tumor were not taken into consideration.
The present study was mostly focused on the search of predictive factors for advanced stage disease and for recurrence, which should alert the gynecological oncologist in proceeding to a complete surgical staging. Indeed, in many cases, the initial operation is a laparoscopic cystectomy, that has been shown to be significantly associated with higher recurrence rates.17 Furthermore, the results of a recently published meta-analysis on serous borderline tumors showed that complete surgical staging significantly reduces the risk of recurrence.18
Our results agree with those observed by other investigators.19–21 Atypical proliferative serous tumors and non-invasive low-grade serous carcinomas occurred in the same age range, but as compared with atypical proliferative serous tumors, micropapillary tumors were more often bilateral with an exophytic pattern of growth and with microinvasive carcinoma. Even though in the above studies microinvasion was not found to worsen prognosis, in our series it was statistically related to recurrence. This finding agrees with the results of Buttin et al.22 On the other hand, there was no such relation for our cases reclassified as ‘microinvasive carcinomas’, a finding probably related to the small number of cases. Nevertheless, all cases featuring both microinvasion and invasive implants had the type of microinvasion designated as ‘microinvasive carcinoma’, either exclusively or mixed with the conventional type of microinvasion. The previous remark underlines the necessity of using the latest WHO classification in the evaluation of microinvasion in serous borderline tumors, since the different types indicate distinct biological behavior of the disease. Indeed, in our cohort, when only single or small clusters of eosinophilic cells were encountered within the stroma of the tumor, no statistical relation to the presence of invasive or non-invasive implants was found.
Several conclusions can be drawn when considering the relation of microinvasion or microinvasive carcinoma to lymph node involvement, since no statistically significant difference was observed between tumors with microinvasion, microinvasive carcinomas and non-microinvasive tumors, in either atypical proliferative serous tumors or non-invasive low-grade serous carcinomas. Moreover, as already mentioned, lymph node involvement was not found to be significantly associated with risk of recurrence. Therefore, we would agree with the trend towards omitting lymphadenectomy in patients with serous borderline tumors, specifically when considering the increased surgical risk and the prolonged operation time. Of course, due to lack of consensus on the management of serous borderline tumors, further studies with extended follow-up periods should be performed to draw firm conclusions, since recent data show that lymphadenectomy was found to be related to improved progression-free survival.23
Among our findings, we should emphasize the absence of statistical significance in the rate of recurrences between atypical proliferative serous tumors and non-invasive low-grade serous carcinomas. Considering survival, only three women with invasive implants died of disease. When analyzing these cases, two of these women had an atypical proliferative serous tumor with an exophytic growth and ‘minimal’ micropapillary pattern. The aggressive behavior encountered in those cases might imply that such micropapillary areas, even ‘minimal’, when involving the ovarian surface, can be the source of invasive implants, hence worsening the biological behavior of the tumor. Of course, before drawing any conclusions, our observation needs to be further validated through large retrospective studies.
Finally, in our group of tumors, the prevalence of endosalpingiosis was 32.8% and this percentage is consistent with other published data.24 Despite the absence of a statistical relation between endosalpingiosis and the other clinicopathological parameters, we believe that the relatively elevated rate of recurrence for cases with endosalpingiosis merits further speculation. Specifically, the development of serous tumors from areas of endosalpingiosis has already been described.25 Recently, an immunophenotypic similarity identified between endosalpingiosis and the normal fallopian tube epithelium supports the hypothesis that serous neoplasia arises from tubal epithelium and not from the ovarian surface epithelium.25 Since in our group of tumors, endosalpingiosis, although of no statistical significance, was related to an elevated rate of recurrence, we can make the hypothesis that some recurrences might not be related to the initial ovarian tumor, but they could probably represent neoplastic transformation, occurring in areas of endosalpingiosis.
Our study has limitations and the results should be interpreted with caution. First, the number of lymphadenectomies performed is relatively small. Yet, since omitting lymphadenectomy is one of the choices in the management of ovarian serous borderline tumors, we believe that our sample could be considered as representative. Second, even though the median overall survival of our cases was over 100 months, it is well-established that recurrences of borderline tumor can occur 10–15 years or even longer after the initial diagnosis. Therefore, the recurrence rate might be underestimated.
The results of our study underline the statistically significant prevalence of bilaterality, exophytic growth, microinvasive carcinoma, and invasive implants in non-invasive low-grade serous carcinomas as opposed to atypical proliferative serous tumors. Recurrences were only related to exophytic tumors, to microinvasion, and to the presence of implants, either invasive or non-invasive. Nevertheless, in multivariate analysis, only microinvasion and invasive implants were found to be a risk factor for recurrence. Lymph node involvement and endosalpingiosis were neither related to the type of tumor nor to recurrence. Women who eventually succumbed to the disease were only those with invasive implants, irrespective of tumor sub-type, justifying their newly proposed classification as low-grade serous peritoneal carcinomas. The indolent nature of ‘minimal’ micropapillary areas encountered in atypical proliferative serous tumors is questioned and should be further studied.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned, externally peer reviewed.