Incomplete Staging in Apparent Early Stage Ovarian Cancer; Guidelines Versus Daily Practice

Background: Incomplete surgical staging of patients with early stage epithelial ovarian cancer (EOC) has been reported in up to 85% of cases, when based on the International Federation of Obstetrics and Gynaecology (FIGO) staging procedure. The aim of the present retrospective study was to clarify the reasons for incomplete staging. Methods: The PRISMA (Prevention Recovery Information System for Monitoring and Analysis) technique was used to evaluate cases with FIGO I-IIa EOC based on incomplete staging from ve gynecologic oncologic center hospitals in the Netherlands in the period 2010-2014. Results: Fifty cases with an incomplete surgical staging of EOC according to national guidelines were included. The most common reasons for incomplete staging were insucient random biopsies of the peritoneum (n=34, 68%), and less than ten lymph nodes being resected and/or found at pathology (n=16, 32%). The most mentioned reasons for not performing biopsies were forgetting to do so, and believing that after careful inspection and palpation, taking biopsies is irrelevant and/or already are being taken while performing a hysterectomy (peritoneum of cul-de-sac, bladder). The value of contralateral pelvic lymph node dissection in case of a unilateral ovarian malignancy was also doubted, inuencing the number of lymph nodes resected. Conclusions: The most important reasons for incomplete staging in EOC are, besides omitting elements by accident, questioning the importance of obligatory elements of the staging procedure. A structured list of staging steps during surgery and more evidence based consensus concerning these obligatory elements might increase the number of complete staging procedures in EOC. oophorectomy, and biopsies and following study: age; histological subtype; tumor stage; differentiation grade; inspection and palpation of all peritoneal surfaces; biopsies of any suspect lesions for metastases; biopsies or removal of any adhesions surrounding the primary tumor; peritoneal washing or ascites; infracolic omentectomy or biopsy; biopsies of peritoneum of the bladder, cul-de-sac, pelvic sidewalls, ovarian fossa, right and left paracolic gutter, and right hemi-diaphragm; number of lymph nodes removed; regions of lymph nodes removed (high para-aortal / paracaval, ipsilateral pelvic, contralateral pelvic). All by or under


Background
In newly diagnosed clinical early stage epithelial ovarian cancer (EOC), the aim of surgery is to resect the tumor combined with an adequate staging procedure. This will provide prognostic information by means of the de nite stage and will de ne whether adjuvant chemotherapy is needed or can be safely omitted.
(1) According to the International Federation of Obstetrics and Gynaecology (FIGO) a staging procedure should include a peritoneal washing or a sample of ascites, hysterectomy and bilateral oophorectomy, infracolic omentectomy (complete, partial or biopsies), numerous peritoneal biopsies from several locations at risk for tumor implantation, pelvic and para-aortic lymphadenectomy and biopsies of suspected lesions and adhesions. (2) The Dutch guideline has incorporated these staging elements with the adjustment that overall at least 10 lymph nodes should be resected from speci c locations in the pelvic and paraaortic area .
Despite the FIGO recommendation, incomplete surgical staging has been reported in up to 85% of cases in different studies. (3)(4)(5)(6) Studies examining the causes of incomplete staging are scarce and often based on assumptions such as lack of specialized skills and/or knowledge of the gynecologist. (3,5,6), The Prevention Recovery Information System for Monitoring and Analysis (PRISMA) was originally developed to manage human error in the chemical process industry, but has also been applied in healthcare (PRISMA-Medical). (7,8,9 ) The main goal of PRISMA is to build a quantitative database of incidents and process deviations, in order to facilitate the development and evaluation of system-based preventive strategies. The PRISMA-Medical method consists of a causal tree incident description followed by a classi cation of root causes. Causal trees are used because nearly all incidents have more than one cause. By continuing to ask 'why' of each event (beginning with the top event), a structure of causes and consequences arises, until the root causes are identi ed at the bottom of the tree. These root causes are subsequently classi ed by linking them to one of the categories. As both active failures (human error) and latent conditions (technical and organizational failures) surrounding incidents are systematically considered with the PRISMA-Medical method, the results of this analysis can be used to provide a more realistic view of daily practice (8). In this study, we used PRISMA to study the causes of incomplete staging in EOC.

Methods
Five hospitals participated in the study and all hospitals ful lled the Dutch criteria to be a gynecologic oncology center within their region among which the performance of 20 or more ovarian cancer debulking procedures each year. Patients with EOC FIGO stage I -IIa undergoing a staging procedure in the period from 2010 until 2014 were included if one of the following incidences had occurred: no sampling of ascites or peritoneal washing, less than ve peritoneal biopsies resection of less than ten lymph nodes; no high para-aortal, ipsi-and/or contralateral pelvic lymph nodes; no omentum biopsy or resection.
Patients in whom a staging procedure was waived in advance (i.e. because of age, comorbidities) were excluded.
Cases were selected retrospectively from the Netherlands Cancer Registry (NCR). The NCR is maintained by the Netherlands Comprehensive Cancer Organization and contains data of all cancer patients in the Netherlands. The NCR, which reached full national coverage in 1989, relies on noti cation of all newly diagnosed malignancies via the automated nationwide pathology archive. Trained registrars use standardized forms to collect patient information from medical records, and regular consistency checks are performed to ensure the quality of the data held in the NCR. The following data were examined in this study: age; histological subtype; tumor stage; differentiation grade; inspection and palpation of all peritoneal surfaces; biopsies of any suspect lesions for metastases; biopsies or removal of any adhesions surrounding the primary tumor; peritoneal washing or ascites; infracolic omentectomy or biopsy; biopsies of peritoneum of the bladder, cul-de-sac, pelvic sidewalls, ovarian fossa, right and left paracolic gutter, and right hemi-diaphragm; number of lymph nodes removed; regions of lymph nodes removed (high para-aortal / paracaval, ipsilateral pelvic, contralateral pelvic). All surgery was performed by or under supervision of quali ed gynecologic oncologists.
The PRISMA commission consisted of eight gynecologists, a member of the NCR (MA), and a medical researcher. All included cases were analyzed and discussed with the gynecologic oncologists of each participating hospital. All members participating in the commission signed a non-disclosure agreement.

Results
Study population Figure 1 shows the study population and reasons for exclusion. Fifty-eight patients were selected by the NCR data managers . One selected patient had advanced stage disease at pathology (IIIA1 based on a lymph node metastasis) and was excludedAfter analyzing and discussing the patients with the gynecologists of each participating hospital, seven patients (12%) appeared to have had a complete staging after all, but were interpreted differently during selection by the NCR data managers. In two of these cases the location of the aortocaval lymph node dissection was not well described, and in ve cases the gynecologist used anatomical terms describing the location, that were unfamiliar to the data managers. Therewith, 50 cases remained for inclusion in this study.
Most patients (n=34, 68%) underwent one operation, all lapartomies with frozen section. In 15 patients (30%) a second operation was performed (of which 10 patients by laparoscopy), and in one patient (2%) despite three surgical procedures (two laparotomies and one laparoscopy) the staging procedure still was incomplete.
Nineteen patients (38%) received adjuvant chemotherapy. In two cases, it was unknown whether adjuvant chemotherapy was given (4%) and one patient refused chemotherapy. For the remaining 29 patients (58%), according to the gynecologic oncologist (and the Dutch guidelines), there was no indication for adjuvant chemotherapy.

Incomplete staging
The number of missing elements was one in 12 patients (24%), two in 17 patients (34%), three in eight patients (16%), four in four patients (8%), ve in four patients (8%), and six or more in the remaining ve patients (10%). Table 1 shows incompleteness of the staging procedure with respect to the different elements. The most frequent reason for incomplete staging was insu cient peritoneal biopsies of the different locations (n=34; 68%). Reasons for not performing certain biopsies were 1) forgetting to do so (n=11), 2) presumption that the biopsy (cul-de-sac and bladder peritoneum) is already part of the hysterectomy (n=5), 3) the opinion of the surgeon that after careful inspection taking biopsies of macroscopically normal peritoneum is irrelevant (n=6), 4) chemotherapy already being indicated, hence complete staging would not be contributory (n=6), 5) technical problems (n=3), 6) comorbidity or malignancy elsewhere (n=2), and 7) the wish of the patient to not receive adjuvant therapy (n=1) (Figure 2).
In eight patients (16%) no lymph nodes were resected, either because it was technically impossible (n=4) or because the outcome of the lymph nodes, according to the gynecologist, had no therapeutic consequences (n=4, Figure 2). In 25 patients (50%) no contralateral pelvic lymph nodes were resected. In eight patients (16%) the number of resected lymph nodes was less than ten. This could be partly explained by the fact that in ve of these patients only lymph nodes were resected on the ipsilateral pelvic region. Furthermore, although according to the gynecologist enough tissue was resected from the paraaortic region, the pathologist did not identify any lymph nodes in one patient (2%). Taking ascites or peritoneal washing was omitted by accident in 12 patients (24%), taking biopsies from adhesions surrounding the tumor was omitted in 14%, a omentum biopsy or omentectomy in 2%, and a biopsy of suspicious lesions in 2% of cases. A hysterectomy or contralateral ovariectomy was not performed in ve cases (10%) to preserve fertility or because of the wish of the patient to not receive adjuvant chemotherapy, and therefore a complete staging was not performed.

Discussion
In this retrospective analysis we found that incomplete staging in EOC, performed by gynecologic oncologists, is most often due to insu cient biopsies of different locations of the peritoneum (n=34; 68%), and less than ten lymph nodes being resected and/or found at pathology (n=16; 32%). The most mentioned reasons for not performing biopsies were forgetting to do so, and believing that, after careful inspection and palpation, taking blind biopsies is irrelevant and/or biopsies already are being taken while performing a hysterectomy (peritoneum of the cul-de-sac, broad ligament and bladder). Forgetting to take blind biopsies was registered in 11 cases and there can be little excuse to justify such negligence.
Previous studies, mainly focused on patients treated before 2000, report incidences of incomplete staging up to 85%. (3,5,6,10) In the majority of these patients no or inadequate lymph node resection was performed. (3), but also no or an insu cient number of random peritoneal biopsies was frequently found. These studies conclude that lack of specialized skills and/or knowledge of the gynaecologist might be a possible explanation for incomplete staging. (3,5,6,11) However, in the present study only gynaecologic oncological center hospitals were included, (of which one can assume that they do have the skills and knowledge that is required), taking blind biopsies was simply forgotten in about one third (11/34) of the incomplete biopsy cases. A standardized surgical report, and check lists with all parts of the staging procedure according to the FIGO guidelines available in the operation rooms, might be a solution to forgetting parts of the procedure.
The assumption that taking blind biopsies is unnecessary is a different matter. In a recent review on surgical staging of EOC the average yield of positive blind biopsies of the peritoneum in a number of studies was around 8%. (12) It has also been argued that upstaging of EOC on the basis of positive blind biopsies is lower but this would be true for all the different staging steps and data on this speci c issue are scarce.  (14) Five of these six patients were upstaged by random biopsies alone. Taking a more general perspective over the debate of percentages it has been argued that random peritoneal biopsies remain indicated in early-stage disease owing to the low morbidity of the procedure and the present possibility of upstaging and altered management. Furthermore, systematic peritoneal biopsies ensure careful palpation and examination of all surfaces. The latter notion is the more important in view of the recent trend to perform staging of EOC laparoscopically, therewith losing the contribution of palpating subperitoneal abnormalities.
Adequate lymph node sampling appears to be an essential part of the staging procedure in clinical early stage EOC. In a review on lymph node metastases, the overall incidence varied from 6.1% to 29.6% (mean 14.2%). (15) Despite its importance, the reported incidence of lymph node staging worldwide varies between 10% and 30% . (16) In a recent report on lymph node staging in The Netherlands, the incidence of lymph node dissection improved from 26% in 2000 to 67% in 2012. Moreover, the percentage of patients from whom 10 or more lymph nodes were removed also increased during the study period (from 2.3% to 47.6%). (17) In the present study, in 16 patients with an incomplete staging procedure (32%), either no lymph nodes were resected or the number of lymph nodes identi ed was below ten. Various reasons for this were mentioned by the gynecologists. First of all, in ve patients no lymph nodes were resected because it had no therapeutic consequences according to the gynecologist (already indication for adjuvant chemotherapy). In another four patients it was mentioned that the procedure was technically impossible, mainly related to intense intra-abdominal adhesions, extensive blood loss and/or obesity. Secondly, although according to the gynecologist enough tissue at the different pelvic and para-aortal locations were sampled, the number of lymph nodes identi ed by the pathologist were less than ten, hence insu cient according to the Dutch guideline. This could be partly explained by the fact that in ve patients pelvic lymph nodes were obtained only at the ipsilateral side. In 25 patients (50%), including these ve patients, the value of a contralateral pelvic lymphadenectomy in case of a unilateral ovarian tumor was questioned and therefore omitted. In the already mentioned review examining the incidence and locations of lymph node metastases in early stage EOC, 20% of all lymph node metastases were located at the pelvic regions only. Only contralateral lymph node metastases of a unilateral tumour was found in 16% of positive pelvic nodes and 11% of positive paraaortic nodes. (15) These ndings provide a strong argument against an ipsilateral node dissection only.
Besides the resection of not enough fatty tissue containing the lymph nodes by the gynecologist, the pathologist examining the tissue resected may in uence the de nite number of lymph nodes recognized. In a study that observed the impact of nodal retrieval after educating the pathologist, found that extra education about the importance of nodal count signi cantly raised the nodal count at pathology. (18) New pathological methods are being investigated and have shown promising results. Svec et al. showed that re-xing in a lymph node revealing solution containing ethanol, diethyl ether, glacial acetic acid and formalin, increased the number of revealed lymph nodes in colorectal resection specimens. (19) In another study comparing standard formalin xation with fat-clearing by acetone in specimens after gastrectomy, more lymph nodes were found using aceton, though this was not statistically signi cant.
More high quality trials are needed to study the optimal technique for retrieving lymph nodes at pathology.
Complete staging has been proven to be an important prognostic factor in EOC. (4,17) There is evidence that, in contrast to incompletely staged patients, adjuvant chemotherapy in completely staged patients does not improve survival and should be omitted. (4,17,20,21) This argument makes a comprehensive and complete surgical staging the more important. Our study demonstrates that the staging quality in daily life practice is still too low. We have analyzed the reasons for this and these data might serve to improve the staging performance of EOC. It should be said that the low percentages of complete surgical staging of EOC according to the existing guidelines might be attered. In our study we found some reasons of the gynecologic oncologists for deliberately omitting certain staging steps in the individual clinical situation, perfectly plausible. In 34 patients the taking of blind peritoneal biopsies was incomplete. Of these, taking the biopsies was considered to have no therapeutic consequences in six patients, there were technical problems in three patients and comorbidity problems or a second malignancy in two. Together they represent one third of the incomplete biopsy group.
In 16 patients the yield of lymph nodes was incomplete with technical problems in four and no therapeutic consequences also in four. To convict these cases as victims of inadequate surgical performance would be unfair. For this reason a plea has been made to consider cases such as these separately from the categories complete and incomplete staging and the term 'considerate adjustment of staging procedure (CASP)' has been proposed. (22) It might be worthwhile to include this category in future studies on staging of EOC. In addition to this, future guidelines should leave more room for tailored procedures such as a stronger indication for surgical lymph node sampling (high grade serous tumors) and possible omission of sampling in case of low grade mucinous histology.

Conclusions
The most common reasons for incomplete staging were insu cient random biopsies of the peritoneum, and less than ten lymph nodes being resected and/or found at pathology. The most mentioned reasons for not performing biopsies were forgetting to do so, or believing that, after careful inspection and palpation, taking biopsies is irrelevant and/or already are being taken while performing a hysterectomy (cul-de-sac, bladder). The value of contralateral pelvic lymph node dissection in case of a unilateral ovarian malignancy was also doubted, in uencing the number of lymph nodes resected. Introducing a third staging category (considered adjustment of staging procedure) might give a more realistic picture of the actual staging performance. A structured list of staging steps to be used during surgery and evidence based consensus of obligatory staging steps might help in achieving a higher rate of complete staging procedures.

Declarations
Ethics approval and consent to participate. The PRISMA commission consisted of eight gynecologists, a member of the NCR (MA), and a medical researcher. All included cases were analyzed and discussed with the gynecologic oncologists of each of the 5 participating hospitals. All members participating in the commission signed a non-disclosure agreement. Ethical approval was obtained in all regional hospitals (Commission of Medical Ethics, February 2018, reference number G17.114) Consent for publication: All authors consented for publication Availability of data and material: Cases were selected retrospectively from the Netherlands Cancer Registry (NCR). The NCR is maintained by the Netherlands Comprehensive Cancer Organization and contains data of all cancer patients in the Netherlands. The NCR, which reached full national coverage in 1989, relies on noti cation of all newly diagnosed malignancies via the automated nationwide pathology archive. Trained registrars use standardized forms to collect patient information from medical records, and regular consistency checks are performed to ensure the quality of the data held in the NCR.
Competing Interest. The authors declare that they have no con ict of interest. Flowchart. Fifty-eight patients with incomplete staging of early epithelial ovarian cancer were included.
One patient was excluded due to pathological con rmed advanced stage ovarian cancer. Complete staging was the reason for exclusion in seven other patients.

Figure 2
Reasons for incomplete staging. Per staging element reasons for omitting the particular element of the staging procedure are described and counted.