Search

ObjectiveHigh grade endometrial stromal sarcoma and undifferentiated uterine sarcomas are associated with a very poor prognosis. Although large surgical resection is the standard of care, the optimal adjuvant strategy remains unclear.MethodsA retrospective analysis of patients with localized high grade endometrial stromal sarcoma and undifferentiated uterine sarcomas (stages I–III) treated in 10 French Sarcoma Group centers was conducted.Results39 patients with localized high grade endometrial stromal sarcoma and undifferentiated uterine sarcomas treated from 2008 to 2016 were included. 24/39 patients (61.5%) were stage I at diagnosis. 38/39 patients underwent surgical resection, with total hysterectomy and bilateral oophorectomy completed in 26/38 (68%). Surgeries were mostly resection complete (R0, 23/38, 60%) and microscopically incomplete resection (R1, 6/38, 16%). 22 patients (58%) underwent postoperative radiotherapy (including brachytherapy in 11 cases), and 11 (29%) underwent adjuvant chemotherapy. After a median follow-up of 33 months (range 2.6–112), 17/39 patients were alive and 21/39 (54%) had relapsed (9 local relapses and 16 metastases). The 3 year and 5 year overall survival rates were 49.8% and 31.1%, respectively, and 3 year and 5 year disease free survival rates were 42.7% and 16.0%, respectively. Median overall survival and disease free survival were 32.7 (95% CI 16.3–49.1) and 23 (4.4–41.6) months, respectively. Medians were, respectively, 46.7 months and 39.0 months among those who underwent adjuvant radiotherapy and 41.0 months and 10.3 months for those who underwent adjuvant chemotherapy. In multivariate analysis, adjuvant radiotherapy was an independent prognostic factor for overall survival (P=0.012) and disease free survival (P=0.036). Chemotherapy, International Federation of Gynecology and Obstetrics I–II stages, and Eastern Cooperative Oncology Group-performance status 0 correlated with improved overall survival (P=0.034, P=0.002, P=0.006), and absence of vascular invasion (P=0.014) was associated with better disease free survival.ConclusionsThe standard treatment of primary localized high grade endometrial stromal sarcoma and undifferentiated uterine sarcomas is total hysterectomy and bilateral oophorectomy. The current study shows that adjuvant radiotherapy and adjuvant chemotherapy appear to improve overall survival. A prospective large study is warranted to validate this therapeutic management.

Background and Objectives Published studies on medical practice assessment often use clinical audit as method. However, clinical audits require considerable resources, in particular human resources, as each medical file has to be reviewed individually. If the number of files is large, this can take time, with furthermore no guarantee that the results will be reproducible. Our objective was to develop algorithms for the automated analysis of compliance of medical decisions with national or regional clinical practice guidelines. Programme Our work was carried out within a financial and organisational partnership between Merck-Serono, the Canceropole CLARA (Canceropole Lyon Auvergne Rhone-Alpes) and the European network of excellence CONTICANET (Connective Tissue Cancer Network). We based our work on medical practice appraisal studies we had performed on the management of patients with non metastatic breast cancer and of patients with sarcoma. We compared the use of human and automated means of assessing compliance with guidelines in a case that was common and one that was rare and required more complex management. We collected data on both types of case in each institution and compared treatment administered to treatment recommended for each patient. When patient management was in line with guidelines, it was considered to comply. When management deviated from guidelines without any medical justification, it was considered not to comply. We determined compliance with guidelines by an observer and by the algorithm and used the kappa coefficient to obtain a reliable and reproducible measure of concordance between the two methods. We then calculated algorithm sensitivity, specifically, positive predictive value (PPV) and megative predictive value (NPV). Results 586 treatments for breast cancer were assessed. kappa for concordance between observer and algorithm was 0.55, indicating that the algorithm is a partial substitute for the observer. Algorithm sensitivity was 46% and specificity was 99%. This means that if the algorithm states that treatment complies with guidelines, the observer need not review the file but can be certain that the answer is correct. This was confirmed by the PPV which was 95.7%. However, the NPV was lower (84.7%). Thus, when the algorithm indicates noncompliance, the observer has to check the decision in the patient9s file. In a study of treatment decisions in patients with sarcoma (gastrointestinal stromal tumour (GIST) or retroperitoneal sarcoma), the kappa coefficient was only 0.27 for GIST and 0.13 for retroperitoneal sarcomas. These preliminary results need to be verified on soft tissue sarcomas which are more common. Discussion and Conclusion Our algorithm for analysing compliance with clinical practice guidelines is a useful tool for optimising medical practice assessment. It can help save time and human resources, improve reproducibility, and guarantee unbiased observer judgement. However, for widespread use of such an algorithm, the disease must be common enough to achieve a reliable sample size and must have a homogeneous morphology and pathology. Contexte Les etudes publiees concernant l9analyse des pratiques medicales suivent souvent la meme methodologie de l9audit clinique. Cette derniere possede quelques lourdeurs organisationnelles, notamment le cout humain, les conformites etant faites dossier par dossier. Leur nombre etant generalement important l9analyse peut s9averer longue et pas toujours reproductible. A partir de ces constats, il nous a semble pertinent de tenter d9automatiser ces modes d9analyse de la conformite des decisions des pratiques medicales aux referentiel des pratiques cliniques par des algorithmes decisionnels. Ceux-ci sont elabores a partir des referentiels nationaux ou regionaux. Cette recherche de nouvel outil a ete soutenue par un multi-partenariat financier et organisationnel par l9industriel Merck-Serono, le Canceropole CLARA (Canceropole Lyon Auvergne Rhone-Alpes) et le reseau d9excellence Europeen CONTICANET (Connective Tissu Cancer Network). Methodologie Nous nous sommes appuyes sur les etudes d9evaluation des pratiques medicales que nous avons realisees concernant la prise en charge des patientes atteintes de cancer du sein non metastatique, et la prise en charge des patients atteints de sarcome. Le but est de comparer l9evaluation de la conformite humaine avec celle que l9on a numerisee, dans le cas d9une pathologie dite «frequente» et dans le cas d9une pathologie rare dont la prise en charge est plus complexe. Pour chacune des deux pathologies et apres avoir collige, dans chaque structures de prise en charge concernees les informations medicales nous avons confronte le traitement de chaque patient aux recommandations afin d9evaluer la conformite de leur prise en charge. Dans le cas ou la prise en charge suivait le referentiel nous l9avons juge conforme. Si une deviance au referentiel etait constatee sans justification medicale la prise en charge etait jugee non-conforme. Nous avons obtenu un taux de concordance aux referentiels etablit a la fois par un observateur et a la fois par un algorithme. Afin d9obtenir une mesure de concordance fiable et reproductible entre les deux methodes, nous avons calcule le coefficient de concordance Kappa. Nous avons ensuite calcule la sensibilite et la specificite. Ces donnees sont egalement accompagnees de la Valeur Predictive Positive (VPP) et la Valeur Predictive Negative (VPN). Resultats 586 sequences therapeutiques ont ete evaluees dans le cadre de la prise en charge du cancer du sein. Il apparait que les algorithmes peuvent en partie suppleer l9observateur puisque nous retrouvons un Kappa de 0.55. La sensibilite et la specificite etant respectivement de 46% et de 99%. Aux vues des resultats, il convient de conclure que lorsque les algorithmes ont code conforme une procedure ou une prise en charge conforme, l9observateur n9a pas besoin de regarder les dossiers il est sur que cette analyse est juste. Ceci est valide par la VPN qui est de 84.7% et la VPP qui a une valeur de 95.7%. En revanche, au vu de la VPN plus basse, lorsque l9algorithme a code non-conforme il est necessaire que l9observateur puisse verifier les decisoins obtenues. Cette demarche a ete reproduite dans le cadre des sarcomes (GIST et retroperitoneaux). Les resultats sont beaucoup plus nuances avec un Kappa de 0.27 pour les GIST et de 0.13 pour les sarcomes retroperitoneaux. Il conviendra de verifier ces premiers resultats concernant cette pathologie, sur les sarcomes des tissus mous qui possedent un effectif beaucoup plus important. Conclusion Il semble interessant d9utiliser cette technique d9analyse de la conformite par un algorithme dans le cadre des evaluations des pratiques medicales. Ainsi que certaines etapes de la methodologie des evaluations des pratiques medicales pouvaient etre supplantees par une automatisation grâce a la programmation d9algorithmes decisionnels. La generalisation de cet outil est correlee a plusieurs parametres. Il faut tout d9abord que la pathologie soit frequente pour obtenir une taille d9echantillon pertinente. La pathologie doit etre homogene dans sa morphologie et son anatomopathologique. Ce type d9outil peut apporter un gain de temps de personnel medical et servir de tres bon controle pour la reproductibilite, et l9impartialite du jugement de l9observateur.

Background Sarcoma represents a morphologically heterogeneous group of malignant tumours, accounting for 1% of malignancies. Given the rarity of cases and the multiplicity of histological types, the diagnostic is difficult to perform. Nevertheless, accurate initial histological diagnosis is essential for adequate management. Over a 1-year period, we conducted a prospective and exhaustive study of the medical management of patients with sarcoma to identify determinants of medical practices and to realise a centralised and systematic review of all cases diagnosed within the Rhone-Alpes (RA) region. The goal of this work was to assess the impact of a systematic pathological review on management of patients. Design All suspicions of sarcoma arising in patients diagnosed in the RA region were collected over a 1-year period. Firstly, the initial diagnosis was compared with the corresponding expert opinion. The concordance between the two diagnoses was assessed according to three defined levels as follows: (0) absolute mismatch when the difference observed between the two diagnoses was major for example, sarcoma vs benign soft tissue tumour or non-sarcoma (melanoma, carcinoma, lymphoma), or misidentification of the histological type of sarcoma potentially resulting in important therapeutic impact (eg GIST vs non GIST); (1) partial concordance when the inconsistency concerned the grade or the sarcoma subtype (eg, dedifferentiated liposarcoma vs myxoid round cell liposarcoma), and (2) absolute concordance when the two diagnoses were identical. Second, cases were separated into two main groups: cases sent for the systematic review (SR) induced by this study and cases addressed for the study and also for a second opinion due to the difficulties to propose a diagnosis (SO). Within each group, the similarity of the two diagnoses was assessed using the same methodology with the 3 defined levels previously described. Results 366 cases were analysed. Concordance analysis showed 19% of absolute mismatch (70/366), 27% of partial concordance (97/366) and 54% of absolute concordance (199/366). Among the absolute mismatches, 8% (30/366) were initially diagnosed as benign soft tissue tumour or nonsarcoma tumour and were sarcoma after expert review, 3% (10/366) were initially diagnosed as sarcoma and were not sarcoma after expert review and 8% (30/366) were misidentification of the histological type of sarcoma. Among partial concordances 8% (30/366) were related to histological subtypes and 19% (67/366) were related only to histological grade. SO represented 51% of the cases (188/366) and SR the remaining 49% (178/366). Within the SR group, the three types of concordance rates observed were: 9% (17/178) of absolute mismatch, 25% (44/178) of partial concordance, and 66% (117/178) of absolute concordance. Within the SO group, the absolute mismatch rate reached 28% (53/188), and there was 28% (53/188) of partial concordance and 44% (82/188) of total concordance. Conclusions 46% of first histological diagnoses were modified with second exert opinion. 27% of cases were concerned by a difference in histological type or subtype and 19% were absolute mismatch resulting in therapeutic impact. The high rate of SO (51% in this study) shows that pathologists are aware of the difficulties they face to make reliable diagnosis in this complex pathology. In this group, the high rate of absolute mismatch (28%) demonstrates that an expert SO is justified. The 9% absolute mismatch rate within a group of patients that would not have benefited from any review without this study (SR) shows the necessity to extend the expert process and to continue pathologists training. Systematic SO improves the quality of diagnosis and clearly benefits patients. Even if more costly in short term, optimal therapeutic management eventually results in reduced overall health care spending. Contexte Les sarcomes representent un groupe de tumeurs tres heterogenes, comptant pour 1% de l9ensemble des tumeurs malignes. Etant donne la rarete des cas et la multiplicite des types histologiques, le diagnostic histologique est difficile a etablir. Neanmoins, un diagnostic histologique initial precis est essentiel pour une prise en charge medicale adequate. De ce fait, nous avons conduit pendant un an une etude prospective et exhaustive de la prise en charge medicale des patients atteints de sarcome dans le but d9identifier les determinants des pratiques medicales et de realiser une relecture systematique des sarcomes diagnostiques en region Rhone-Alpes. L9objectif de ce travail etait d9estimer l9impact d9une relecture pathologique ssytematique sur la prise en charge des patients. Programme Toutes les suspicions de sarcomes diagnostiquees en region Rhone-Alpes ont ete collectees durant une periode de un an. En premier lieu, le diagnostic initial a ete compare avec celui de la relecture par le pathologiste expert. La concordance entre les deux diagnostics a ete etablie selon trois niveaux, definis de la facon suivante: (0) discordance totale lorsque la difference observee entre les deux diagnostics etait majeure, par exemple un diagnostic de sarcome vs un diagnostic de tumeur benigne ou de tumeur non sarcomateuse, ou une erreur dans le type histologique ayant un impact therapeutique important (par exemple GIST vs non GIST); (1) concordance partielle lorsque la difference entre les deux diagnostics concernait le grade ou le sous-type histologique (par exemple liposarcome dedifferencie vs liposarcome myxoide), et (2) concordance totale lorsque les deux diagnostics etaient identiques. En second lieu, les sarcomes ont ete separes en deux groupes: les cas ayant ete envoyes pour la relecture systematique conduite dans cette etude et les cas ayant ete envoyes spontanement lors d9une demande de second avis. Pour chacun des deux groupes, la concordance entre les diagnostics a ete etablie selon la meme methodologie. Resultats 366 suspicions de sarcome ont ete analysees. L9etude de la concordance des diagnostics a montre une discordance totale pour 19% des cas (70/366), une concordance partielle pour 27% des cas (97/366) et une concordance totale pour 54% des cas (199/366). Parmi les discordances totales, 8% (30/366) des cas ont ete diagnostiques initialement comme des tumeurs benignes ou des tumeurs non sarcomateuses alors que le diagnostic de sarcome a ete avere apres la relecture, 3% des cas (10/366) ont ete diagnostiques initialement comme des sarcomes et ne l9etaient pas apres la relecture par l9expert et 8% (30/366) comprenaient une erreur dans le type histologique du sarcome. Parmi les concordances partielles, 8% (30/366) etaient lies au sous-type histologique et 19% (67/366) etaient lies uniquement au grade histologique. Les demandes de second avis spontanes representaient 61% des cas (188/366) et la relecture systematique les 49% restants (178/366). Dans ce dernier groupe les taux de concordance observes etaient: 9% (17/178) de discordance totale, 25% (44/178) de concordance partielle et 66% (117/178) de concordance totale. Dans le groupe des demandes de second avis le taux de discordance totale atteignait 28% (53/188) le taux de concordance partielle 28% (53/188) et le taux de concordance totale 44% (82/188). Discussion 46% des dignostiques histologiques initiaux ont ete modifies suite a la relecture histologique. 27% des cas etaient concernes par une difference dans le type ou le sous-type histologique et 19% etaient des discordances totales ayant un impact therapeutique. Le taux eleve de demandes spontanees de second avis (51%) montre que les pathologistes sont conscients des difficultes qu9ils rencontrent pour etablir des diagnostics fiables dans cette pathologie complexe. Dans ce groupe le taux eleve de discordance totale demontre que la relecture histologique est justifiee. Le taux de discordance totale de 9% dans le groupe de patients qui n9aurait pas beneficie de relecture sans cette etude montre la necessite d9etendre la relecture par un expert et de continuer la formation des pathologistes. La relecture systematique ameliore la qualite du diagnostic et beneficie clairement au patient. Meme si elle est plus couteuse a court terme, une prise en charge therapeutique optimale conduit a la reduction des depenses globales de sante.

Background and Objectives A cancer network was created by Lyon University and by public and private general hospitals of the Rhone-Alpes region in 1995 to improve the quality of cancer care and prescribing practices. After an assessment of the impact of the implementation of criteria derived from clinical practice guidelines (CPGs) relating to the management of patients with breast cancer, significant changes were observed in the network whereas none were observed in a control region with no network and no criteria for CPG implementation (Ray-Coquard et al , 2002).1 An audit carried out 3 years after this first assessment revealed that compliance with CPG criteria had been maintained within the network (Ray-Coquard et al , 2005).2 However, a retrospective study (2001) in the same region gave very different results for the management of rare cancers such as sarcomas (Ray-Coquard et al , 2004).3 Our objective was to perform a prospective evaluation of the management of patients with sarcomas and compare compliance with CPGs in 2005 to that recorded in 2001 and to that obtained for breast cancer. Programme We evaluated prospectively medical practices in the management of patients with sarcomas over a 1-year period (2005). A meeting on the findings of the 2001 retrospective study was held in 2002. Its purpose was to act as a reminder on guidelines for the management of patients with sarcoma. Three years later, all the sarcoma cases included in the study, from the first symptom to relapse or death, had been evaluated. Results 691 cases of suspected sarcoma were reported by the pathologists and 439 were included in our evaluation. Exclusions were due to an incorrect diagnosis, elapsed inclusion date, and relapses. The absence of guidelines for certain subtypes of sarcoma (e.g. Kaposi) meant that 410 cases could be included in the final analysis. Overall compliance was 36% (148/410) in 2005 compared to 32% in 2001. The breakdown of compliance (2005 vs 2001) was as follows: 86% vs 75% for pretherapy, 95% vs 85% for histology, 71% vs 48% for surgery, 86% vs 94% for chemotherapy, 97% vs 90% for radiotherapy and 93% vs 79% for follow-up. Compliance rate remained stable between 2001 and 2005. It was not as good as for breast cancer. Discussion and Conclusion The use of criteria derived from CPGs led to sustained changes in medical practice in our cancer network for the management of breast cancer. Compliance with criteria was a result of intensive implementation and clinician involvement in the drafting of the criteria. However, compliance with CPGs did not increase in the case of the management of sarcomas. It would seem that implementation of CPGs and information meetings are less effective in improving care of patients with rare rather than frequent tumours. Contexte et objectifs En 1995, l9universite et des centres hospitaliers publiques ou prives d9une region francaise ont cree un reseau de cancerologie afin d9ameliorer la qualite des soins et de rationnaliser les prescriptions medicales dans la prise en charge des soins en cancerologie. Dans un premier temps nous avons evalue l9impact de la mise en place d9un referentiel sur les pratiques cliniques dans la prise en charge des patientes atteintes de cancer du sein. Des changements significatifs ont ete observes au sein du reseau alors qu9aucun changement n9etait observe dans une region controle sans reseau de cancerologie et sans implementation de referentiels cliniques (Ray-Coquard 2002). Dans un deuxieme temps et 3 ans apres la premiere evaluation,nous avons evalue la persistance dans le temps des conformites aux referentiels au travers d9un nouvel audit medical. Nous avons eu la confirmation dans cette etude qu9un reseau de cancerologie est a meme de maintenir les pratiques medicales conformes aux referentiels (Ray-Coquard 2005). Au vu des resultats tres differents dans une etude retrospective sur la prise en charge de cancers rares comme les sarcomes dans la meme region (Ray-Coquard 2004), nous avons decide d9evaluer pendant 1 an les pratiques medicales dans la prise en charge des sarcomes au sein d9une cohorte prospective et exhaustive de patients concernes par cette pathologie. Et nous les avons compare aux referentiels nationaux. Suite a l9etude menee en 2001, une reunion a ete organisee en 2002 reporter les resultats de l9etude retrospective et rappeler les regles de prise en charge pour ces tumeurs. Ainsi, 3 ans plus tard tous les sarcomes inclus dans cette nouvelle etude evaluant la prise en charge des patients depuis les premiers symptomes a la rechute ou le deces ont ete evalues. Patients et methodes 691 patients presentant une lesion suspecte de sarcome ont ete declares par les anatomopathologistes: 439 ont vraiment ete inclus (le differentiel est du au mauvais diagnostic, date d9inclusion hors delai et patients declares en rechute). En raison de l9absence de recommandations pour certains sous-types (Kaposi, etc…) 410 ont finalement ete inclus dans l9etude. Nous avons lance une etude prospective et exhaustive (afin d9avoir un effectif representatif) dans la region Rhone-Alpes en 2005 pour evaluer la prise en charge des patients atteints de sarcome afin d9evaluer les pratiques medicales dans cette pathologie. Le taux de compliance aux referentiels a ete compare a la premiere evaluation realisee en 2001 ainsi qu9aux resultats obtenus pour le cancer du sein. Resultats Le taux de conformite globale en 2005 etait de 148/410 (36%) contres 32% en 2001. Le taux de compliance etait donc stable entre 2001 et 2005 et ne semble pas aussi interessant que celui du cancer du sein. Conclusion Le programme des referentiels pour la prise en charge des cancers entraine des changements persistants dans les pratiques medicales dans le temps dans notre reseau de cancerologie en terme de conformite aux referentiels pour le cancer du sein. Les referentiels ont ete suivis lorsqu9ils ont une implementation intensivea ete realisee et que les cliniciens ont participe a leur ecriture. En ce qui concerne les sarcomes, l9implementation de referentiel n9a pas augmente le taux de conformite. Pour les cancers rares l9implementation de referentiels et les reunions informatives ne semblent pas plus efficaces pour ameliorer les pratiques medicales contrairement aux tumeurs frequentes.