京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第20240号 / 医博第4199号 / 新制||医||1020(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 川口 義弥, 教授 安達 泰治, 教授 小西 靖彦 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

No description available.

No description available.

Le cancer colorectal est un enjeu majeur de Santé publique. Près de la moitié des patients porteurs d’un cancer colorectal va développer des métastases hépatiques. La chirurgie est le seul traitement potentiellement curatif. Tout doit donc être mis en oeuvre pour que ces patients accèdent à un geste chirurgical. Le volume de foie restant après hépatectomie est un des principaux facteurs limitant en chirurgie hépatique. Des techniques innovantes ont été développées dans l’optique d’une épargne parenchymateuse : l’utilisation conjointe des techniques de destruction focalisée et la chirurgie hépatique en deux temps. La chirurgie hépatique en deux temps est efficace d’un point de vue carcinologique, mais reste une procédure complexe techniquement avec une morbidité non négligeable, notamment à cause des adhérences péri-hépatiques post-opératoires. Ces adhérences sont systématiques après chirurgie hépatique mais peuvent être prévenues par l’utilisation de membranes antiadhérences après la première hépatectomie. Dans cette indication, l’étude de phase II multicentrique, présentée dans ce travail, retrouvait une diminution de l’incidence et de la sévérité de ces adhérences après utilisation de seprafilm®, ce qui facilitait la seconde hépatectomie. Les techniques de destruction focalisée peuvent dans certains cas se substituer à la chirurgie en cas de métastases résécables. Elles permettent également d’augmenter le nombre de patients candidats à une prise en charge à visée curative, lorsqu’elles sont associées à la chirurgie, en cas de métastases non résécables. Ces techniques présentent néanmoins plusieurs inconvénients qui limitent leur utilisation. Les ultrasons focalisés de haute intensité (HIFU) sont une technique récente, non ionisante, de destruction focalisée dont les avantages théoriques sont particulièrement adaptés au traitement des tumeurs hépatiques. La technologie HIFU actuelle repose sur un dispositif de traitement extra-corporel, dont la limite principale est la faible taille des ablations, qui doivent être juxtaposées pour traiter des tumeurs de quelques centimètres, ce qui nécessite des temps de traitement de plusieurs dizaines de minutes. Le développement d’une sonde HIFU per-opératoire, à géométrie torique, a permis d’obtenir des ablations d’environ 7 cm3 en 40 secondes sur un modèle porcin. Nous avons pu montrer, lors de l’étude de phase I-IIa présentée dans ce travail, que ces résultats étaient reproduits chez l’homme sur foie sain destiné à être réséqué. Les résultats positifs de ces deux études prospectives nous ont permis d’envisager une étude de phase III sur la prévention des adhérences péri-hépatiques, la poursuite de l’étude HIFU en ciblant cette fois les métastases hépatiques, et la réalisation d’une étude de phase II sur la résection hépatique assistée par HIFU comme aide à l’hémostase / Colorectal cancer is a major health problem. Almost half of patients will develop liver metastasis. Surgery is the only potentially curative treatment. Everything possible must be done for these patients to perform liver surgery. Remnant liver volume after hepatectomy is the main limit of liver surgery. Innovative techniques have been developed to spare liver parenchyma: concomitant use of focal destruction and two-stage hepatectomy. Two-stage hepatectomy is oncologically effective but is a challenging procedure with high morbidity, in part because of peri-hepatic adhesions. These adhesions occur systematically after liver surgery but can be prevented by the use of anti-adhesion membranes at the end of the first hepatectomy. In this indication, the multicentre phase II study presented herein, showed a decrease in extent and severity of adhesions with use of seprafilm®. It facilitated the dissection and so the second hepatectomy. Techniques involving focal destruction can replace surgery in selected cases of resectable metastases. They also increase the number of patients candidates to curative liver-directed therapy in unresectable metastases. These techniques have however several disadvantages, which limit their use. High intensity focused ultrasound (HIFU) is a recent, non-ionizing, technology of focal destruction. Theoretical advantages make HIFU a promising technique for focal ablation of liver tumours. Current technology is based on extra-corporeal treatment. The main limit is that elementary ablations are small and must be juxtaposed to treat supra-centimetric tumours, resulting in long-time treatment. A new and powerful HIFU device enabling destruction of larger liver volumes (7 cm3 in 40 seconds) has been developed based on toroidal transducers. We showed, in a phase I-IIa study presented herein, that preclinical results could be reproduced on healthy liver of patients undergoing hepatectomy. Positive results of these two prospective studies have allowed to design a phase III trial on prevention of peri-hepatic adhesions, to continue the evaluation of HIFU by targeting liver metastases and by assisting liver resection, in a phase II study, as a sealing device

Introdução: A setorectomia lateral esquerda (SLE) é uma das ressecções hepáticas em que a cirurgia minimamente invasiva tem maior aplicabilidade, sendo considerada um procedimento exequível e passível de padronização técnica. Apesar da crescente aceitação da SLE laparoscópica, o corpo de evidência que suporta a segurança do método, bem como os potenciais benefícios em relação à abordagem convencional são, na sua maioria, provenientes de estudos retrospectivos com número limitado de pacientes. Apenas recentemente, estudos populacionais com grande número de pacientes e estudos randomizados foram publicados acerca do tema. No entanto, estes novos dados não foram avaliados adequadamente por meio de revisão sistemática e metanálise. Desta forma, ainda há carência de evidência de melhor qualidade que possa comprovar os reais benefícios do método. Método: Foi realizada revisão sistemática da literatura nas bases de dados Medline, EMBASE, Cochrane Library Central e LILACS. A data limite de busca foi 31 de dezembro de 2017. Foram incluídos estudos comparativos (retrospectivos e prospectivos) que comparassem os resultados perioperatórios de pacientes submetidos à SLE minimamente invasiva e convencional. Foram excluídos estudos com pacientes submetidos à SLE para doação de fígado. Os desfechos estudados foram taxa de conversão, tempo operatório, perda sanguínea, taxa de transfusão, tempo de internação, morbidade e mortalidade perioperatória (até 90 dias). Resultados: Foram avaliados inicialmente 2.838 artigos, sendo 23 estudos incluídos na metanálise (21 observacionais e 2 randomizados), totalizando 3.415 pacientes. A taxa de conversão foi de 7,4%. Os pacientes submetidos à SLE laparoscópica apresentaram menor perda sanguínea (diferença de médias [DM]=-119,81ml; IC95% -127,90 a -111,72; p < 0,00001; I2=32%; n=618), menor taxa de transfusão (4,1% vs. 10,1%; diferença de risco [DR]=-0,06; IC95% -0,08 a -0,05; p < 0,00001; I2=13%; n=2.968) e menor tempo de internação (DM=-2,02 dias; IC95% -2,15 a -1,89; p < 0,00001; I2=77%; n=3.160). Observou-se ainda diminuição marginal na frequência de complicações (21,4% vs. 27,5%; DR=-0,03; IC95% -0,06 a 0,00; p=0,05; I2=0%; n=3.268) e mortalidade perioperatória (0,3% vs. 1,5%; DR=-0,01; IC95% -0,02 a 0,00; p=0,01; I2=0%, n=3.332) nos pacientes submetidos à cirurgia minimamente invasiva. Não houve diferença quanto ao tempo operatório, complicações biliares, cardíacas e pulmonares entre os grupos. Conclusão: As evidências atuais suportam a abordagem minimamente invasiva como via de acesso padrão para SLE, sendo um procedimento exequível, seguro e associado à redução da perda sanguínea, menor taxa de transfusão e menor tempo de internação hospitalar / Introduction: Left lateral sectionectomy (LLS) is considered a safe and straightforward procedure that can be standardized. Despite increasingly performed, studies comparing laparoscopic and open LLS are mostly retrospective with underpowered sample size. Recently, population-based studies and prospective trials brought new data, however this new evidence was not evaluated in a systematic review and meta-analysis. Therefore, a high quality evidence of the actual benefits of minimally invasive LLS is still lacking. Methods: A systematic review until 31st December 2017 was performed in Medline, EMBASE, Cochrane Library Central and LILACS. Randomized and observational studies comparing perioperative results between laparoscopic and open LLS were included. Studies with patients submitted to LLS for living donation were excluded. Treatment outcomes, including conversion rate, estimated blood loss, transfusion rate, operative time, hospital stay, morbidity and mortality (up to 90th day) were evaluated. Results: A total of 2,838 articles were initially evaluated, and 23 studies (21 observational and 2 randomized trials) were included in the meta-analysis (3,415 patients). The conversion rate was 7.4%. Patients submitted to laparoscopic LLS showed lower blood loss (mean difference [MD]=-119.81ml, 95%CI -127.90 to -111.72, P < 0.00001, I2=32%, N=618), lower transfusion rate (4.1% vs. 10.1%; risk difference [RD]=-0.06, 95%CI -0.08 to -0.05, P < 0.00001, I2=13%, N=2,968), and shorter hospital stay (MD=-2.02 days, 95% CI -2.15 to -1.89, P < 0.00001, I2=77%, N=3,160). Additionally, marginal decrease in overall complications (21.4% vs. 27.5%; RD=-0.03, 95%CI -0.06 to 0.00, P=0.05, I2=0%, N=3,268) and perioperative mortality (0.3% vs. 1.5%; RD=-0.01, 95%CI -0.02 to 0.00, P=0.01, I2=0%; N=3,332) were observed. There was no difference regarding operative time, biliary, cardiac and pulmonary complications between groups. Conclusion: Current evidence supports that laparoscopic LLS is a safe and feasible procedure associated with reduced blood loss, lower transfusion rate and shorter hospital stay. Laparoscopic should be recommended as the gold-standard approach for LLS

INTRODUÇÃO: A ressecção hepática é o único tratamento potencialmente curativo para as metástases hepáticas de câncer colorretal (MHCCR), porém com taxas de recidiva entre 60 e 70%. Desta forma, faz-se necessário um melhor entendimento das vias de disseminação e recidiva da doença. Os brotamentos tumorais (BT) e grupamentos pouco diferenciados (GPD), relacionados à transição epitélio-mesenquimal, são fatores prognósticos para o câncer colorretal; entretanto sua presença e importância nas MHCCR ainda não estão estabelecidas. Os objetivos deste trabalho foram avaliar a presença dos BT e GPD nas MHCCR, determinar sua importância prognóstica e sua relação com outros fatores patológicos conhecidos. MÉTODOS: Foram avaliados, retrospectivamente, 229 pacientes submetidos a ressecção de MHCCR com intuito curativo, entre janeiro de 2004 e junho de 2014. Nos espécimes cirúrgicos das MHCCR ressecadas, foi realizada análise anatomopatológica através de lâminas coradas em hematoxilina e eosina (HE), para avaliação dos BT, GPD, infiltrado inflamatório peritumoral, presença de pseudocápsula e tipo de borda; e por imuno-histoquímica, por meio de anticorpos anti-AE1/AE (BT e GPD), anti-D2-40 (invasão linfática) e anti-CD34 (invasão venosa portal). Também foram analisadas as variáveis clínicas relacionadas ao prognóstico. RESULTADOS: O seguimento médio após a hepatectomia foi de 43 meses. As taxas de sobrevida global e livre de doença em 1, 3 e 5 anos foram de, respectivamente, 94,1%, 66,7% e 45,5% e 53,6%, 31,5% e 29,6%. Os BT estiveram presentes em 61,1% dos pacientes na avaliação pelo AE1/AE3 e em 48,9% pelo HE, enquanto os GPD estiveram presentes em 42,8% dos pacientes na avaliação pelo AE1/AE3 e em 49,3% pelo HE. Na análise univariada, os BT e os GPD grau 3 ( > 9 GPD) mostraram significância prognóstica, tanto na avaliação pelo AE1/AE3 quanto pelo HE. Na análise múltipla, os fatores independentes para sobrevida global foram: a presença de GPD grau 3 (pelo HE), presença de pseudocápsula, invasão venosa portal e presença de 4 ou mais nódulos. Os fatores independentes para sobrevida livre de doença foram: GPD grau 3 (pelo HE), quimioterapia prévia, presença de 4 ou mais nódulos, infiltrado inflamatório peritumoral ausente ou leve, invasão venosa portal e borda infiltrativa. Os BT e GPD não estiveram associados à recidiva hepática. Os BT foram associados a: GPD, MHCCR sincrônicas, tumores de até 5 cm, ausência de pseudocápsula, borda infiltrativa, e presença de invasão venosa portal. Os GPD foram associados a: BT, infiltrado inflamatório peritumoral ausente ou leve, ausência de pseudocápsula e borda infiltrativa. CONCLUSÕES: Os BT e os GPD são frequentes nas MHCCR e ambos, na análise univariada, são fatores de pior prognóstico na sobrevida global e livre de doença, estando associados à recidiva extra-hepática. A presença de GPD grau 3 avaliada pelo HE é fator prognóstico independente na sobrevida global e livre de doença, sugerindo que este é um importante mecanismo de disseminação tumoral na MHCCR / INTRODUCTION: The only potentially curative treatment for colorectal liver metastases is surgical resection. However, about 60 to 70% of patients will recur, showing the necessity of a better knowledge regarding spread and disease recurrence pathways. Tumor budding and poorly differentiated clusters, markers related to epithelial-mesenchymal transition, are prognostic factors for colorectal cancer, but their presence and significance in colorectal liver metastases is not yet established. The aims of this study were to evaluate the presence of tumor budding and poorly differentiated clusters in colorectal liver metastases, to determine their prognostic value and to relate them to other known pathological factors. METHODS: A total of 229 patients that underwent hepatic resection for colorectal liver metastases between January 2004 and June 2014 were retrospectively evaluated. Resected specimens of colorectal liver metastases were submitted to pathological evaluation by hematoxilin and eosin staining, to analyze tumor budding, poorly differentiated clusters, peritumoral inflammatory infiltrate, presence of tumor pseudocapsule and tumor growth pattern; and by immunohistochemical staining with antibodies anti- AE1/AE3 (tumor budding and poorly differentiated clusters), anti-D2-40 (lymphatic invasion) and anti-CD-34 (portal vein invasion). Clinical variables related to prognosis were also evaluated. RESULTS: Mean follow up after hepatectomy was 43 months. Overall and disease-free survival rates at 1, 3 and 5 years were 94.1%, 66.7% and 45.5% and 53.6%, 31.5% and 29.6%, respectively. Tumor budding was present in 61.1% of patients in the evaluation by AE1/AE3 staining and in 48.9% by hematoxilin and eosin, while poorly differentiated clusters were present in 42.8% of patients by AE1/AE3 staining and 49.3% by hematoxilin and eosin. At univariate analysis, tumor budding and poorly differentiated clusters grade 3 ( > 9 poorly differentiated clusters) by AE1/AE3 and hematoxilin and eosin staining showed prognostic significance. On multiple analysis, independent factors for overall survival were: presence of poorly differentiated clusters grade 3 (by hematoxilin and eosin), presence of tumor pseudocapsule, portal vein invasion and presence of 4 or more nodules. Independent factors for disease-free survival were: poorly differentiated clusters grade 3 (by hematoxilin and eosin), preoperative chemotherapy, portal vein invasion, presence of 4 or more nodules, none/mild peritumoral inflammatory infiltrate and infiltrative tumor border. Tumor budding and poorly differentiated clusters were not related to hepatic recurrence. Tumor budding was associated with: presence of poorly differentiated clusters, synchronous colorectal liver metastases, tumor size up to 5 cm, absence of tumor pseudocapsule, infiltrative tumor border and presence of portal vein invasion. Poorly differentiated clusters were associated with: presence of tumor budding, none/mild peritumoral inflammatory infiltrate, absence of tumor pseudocapsule and infiltrative tumor border. CONCLUSIONS: Tumor budding and poorly differentiated clusters are frequent in colorectal liver metastases and, on univariate analysis, are prognostic factors for overall and disease-free survival, being associated to extrahepatic recurrence. Presence of poorly differentiated clusters grade 3 stained by hematoxilin and eosin is an independent prognostic factor for overall and disease-free survival, suggesting that it is an important spread pathway in colorectal liver metastases

Les données cliniques supportent le concept et offrent l’espoir que la thérapie cellulaire trouvera sa place parmi les stratégies thérapeutiques des pathologies hépatiques. Cependant deux obstacles majeurs limitent l'étendue de son application clinique: la faible disponibilité d’hépatocytes humains de qualité et en quantité importante, et une faible efficacité de greffe conduisant à une survie et une fonctionnalité seulement à court terme. L’objectif de ce travail était de développer des modèles animaux d’insuffisance hépatique après hépatectomie et d’analyser la réponse régénérative après transplantation de progéniteurs hépatiques humains isolés et caractérisés dans 2 laboratoires de recherche (INSERM U1040 Montpellier et laboratoire PEDI UCL Bruxelles), en comparaison à des hépatocytes fraichement isolés.Le premier modèle consistait à réaliser une hépatectomie de 30% chez des souris NOD SCID, associée à l’injection préalable de rétrorsine (blocage de la prolifération cellulaire endogène) et d’injecter dans le même temps directement dans le parenchyme 1 million de cellules progénitrices exprimant constitutivement le gène rapporteur Luciférase. Les résultats ont montré la bonne implantation des cellules jusqu’à 1 mois après l’injection avec une différenciation fonctionnelle des cellules mise en évidence par la sécrétion d’albumine humaine dans le sang circulant des animaux.Le deuxième modèle consistait à réaliser une hépatectomie large de 70% chez des souris immunodéprimées RAG 2-/- γ-/- pour augmenter la souffrance hépatocellulaire et à comparer deux timing d’injection (voie intrasplénique) des cellules progénitrices génétiquement marquées avec la Green Fluorescent Protein. Les résultats ont montré une meilleure tolérance clinique (moins de mortalité) et une plus grande quantité de cellules implantées lorsque l’injection était réalisée 48h après l’hépatectomie. La régénération hépatique endogène était plus importante et plus rapide chez les souris injectées avec les progéniteurs qu’avec les hépatocytes primaires, suggérant un effet paracrine bénéfique de ces cellules.Ces travaux ont mis en évidence la possibilité d’utiliser ces cellules progénitrices comme alternative aux hépatocytes avec des propriétés régénératrices certaines mais soulèvent les problèmes d’implantation de ces cellules qui reste faible dans des foies hépatectomisés remaniés. Définir le meilleur environnement pour favoriser la survie, la fonctionnalité et éventuellement l’intégration effective des cellules transplantées reste une question clé pour avancer dans cette voie.En parallèle de ces travaux de recherche, un projet de recherche clinique de biothérapie a été développé et accepté pour transplanter des hépatocytes frais humains en intrahépatique chez des patients ayant une insuffisance hépatocellulaire terminale (hépatite alcoolique aigue, cirrhose grave, après résection hépatique large). Il devrait débuter fin 2013. / Clinical data support the concept and offer the hope that cell therapy will find its place among the therapeutic strategies in liver diseases. However, two major obstacles limit the scope of its clinical application: the limited availability of human hepatocytes quality and in large quantities, and low efficiency leading to graft survival and only a short-term functionality. The objective of this work was to develop animal models of liver failure after hepatectomy and analyze the regenerative response after transplantation of human hepatic progenitors isolated and characterized in two research laboratories (INSERM U1040 Montpellier laboratory PEDI UCL Brussels) compared to freshly isolated hepatocytes.The first model was to achieve a 30% hepatectomy in mice NOD SCID associated with prior injection retrorsine (blocking of endogenous cellular proliferation) and injected at the same time directly into the parenchyma 1 million progenitor cells constitutively expressing the luciferase reporter gene. The results showed good cell implantation until 1 month after injection with a functional differentiation as evidenced by secretion of human albumin in the circulating blood cells of animals.The second model was to achieve a wide 70% hepatectomy in mice immunocompromised RAG 2 - / - γ-/ - to increase the suffering hepatocellular comparing two injection timing (channel intrasplenically) progenitor cells genetically marked with the Green Fluorescent Protein. The results showed better clinical tolerance (less mortality) and a greater amount of implanted when the injection was performed 48 hours after hepatectomy cells. Endogenous hepatic regeneration was greater and faster in mice injected with the progenitors with primary hepatocytes, suggesting a beneficial paracrine effect of these cells.These studies have highlighted the possibility of using these progenitor cells as an alternative to hepatocytes with regenerative properties but raise some problems implementing these cells remains low in hepatectomized livers reworked. Define the Define the best environment to promote the survival, function and possibly the effective integration of transplanted cells remains a key issue for progress in this direction.In parallel with this research, a clinical research project biotherapy was developed and agreed to transplant human hepatocytes in intrahepatic costs in patients with terminal liver failure (acute alcoholic hepatitis, severe cirrhosis, after extensive liver resection). It should begin in late 2013.

La principale limite en chirurgie hépatobiliaire est représentée par l’insuffisance hépatocellulaire (IHC) posthépatectomie (Hx) ou la dysfonction du greffon (EAD) après transplantation (TH). Peu d’études ont évalué le métabolisme du foie dans son ensemble, du fait du manque de technique utilisable en clinique. La métabolomique HR-MAS-RMN pourrait pallier à ce manque. Le but de cette thèse était d’évaluer l’apport de cette technique en chirurgie hépatobiliaire.En TH (n=42), le profil métabolique (PM) prédisait le risque d’EAD et identifiait le lactate et la phosphocholine comme biomarqueurs permettant d’envisager un matching métabolique. Après Hx majeure (n=45), le PM prédisait la survenue d’un décès par IHC. Ce PM différait du profil cirrhotique en décompensation et était compatible avec celui de système cellulaire prolifératif. Une étude préliminaire montrait que le PM prédisait la récidive à 1 an après hépatectomie. Ce travail montre l’intérêt de la métabolomique par HR MAS RMN pour prédire l’issue d’une Hx ou d’une TH dans un temps compatible avec la clinique. Ces données orientent vers la piste de l’intervention métabolique en chirurgie hépatique. / One of the main limits in liver surgery is the risk of liver failure (LF) after hepatectomy (Hx) or graft dysfunction (EAD) after liver transplantation (LT). Few studies have evaluated global liver metabolism, probably due to the lack of clinically relevant techniques. HR-MAS-NMR metabolomics may fulfill this lack and the goal of this work was to evaluate its capacity to predict early outcomes after hepatectomy and LT. In LT (n=42), metabolic profile predicted EAD and lactate and phosphocholine were potent biomarkers providing means for metabolic matching. In liver biopsies harvested at the end of major Hx (n=45), metabolic profile predicted PHLF. The profile at risk of LF differed from that of decompensated cirrhosis but correlated to that of proliferative multicellular systems. A preliminary study showed that the metabolic profile predicted the risk of liver metastases recurrence at 1 year. This work underlines the potential value of HR-MAS-NMR metabolomics in the prediction of short-term outcomes in liver surgery. It provides clues to be further investigated for future evaluation of metabolic intervention in the field of liver surgery.

Le principal frein au développement de la transplantation hépatique à donneur vivant est le risque de complication et de décès encourus par le donneur. Nous proposons une préparation du donneur par la réalisation d'un rétrécissement (sténose) de 20% de la veine d'un coté de son foie afin de faire grossir ce dernier sans altérer la viabilité du futur greffon. Pour cela, sur 32 porcs, nous avons recherché la plus petite sténose capable de déclencher le maintien du débit portal et étudié les déclenchements de la prolifération cellulaire et de l'atrophie. Des scanners associés à des scintigraphies hépatobiliaires à la mebrofénine ont été réalisés afin d'étudier les changements morphologiques et fonctionnels du foie. Nous démontrons que la sténose de 20% d'un coté du foie déclenche la régénération hépatique de l'autre coté et permet le gain d'une masse hépatique fonctionnelle. Conclusion : Notre pré-conditionnement est capable de préparer le foie d'un patient à l'hépatectomie majeure. / The main hindrance in promoting living donor liver transplantation remains the morbidity and mortality risk for the donor. We propose the realization of a 20% stenosis of the left portal vein (LPV) in order to induce an increase of the functional liver mass without altering the viability of the future graft.Materials and Methods: Thirty-two pigs were included in this program. The hemodynamic study identified the smallest stenosis capable of triggering mechanisms of maintenance of the hepatic blood flow. Cell proliferation and atrophy were studied. Scanners associated with Mebrofenin hepatobiliary scintigraphy were performed to study the morphological and functional changes of the liver.Results: A 20% LPV stenosis trigger liver regeneration in the contralateral lobe inducing a gain in hepatic functional hepatic mass.Conclusion: A 20% LPV stenosis is the effective preconditioning in order to get the remnant liver of living donor ready to take on graft harvesting.

INTRODUÇÃO: A ressecção hepática é o único tratamento potencialmente curativo para as metástases hepáticas de câncer colorretal (MHCCR), porém com taxas de recidiva entre 60 e 70%. Desta forma, faz-se necessário um melhor entendimento das vias de disseminação e recidiva da doença. Os brotamentos tumorais (BT) e grupamentos pouco diferenciados (GPD), relacionados à transição epitélio-mesenquimal, são fatores prognósticos para o câncer colorretal; entretanto sua presença e importância nas MHCCR ainda não estão estabelecidas. Os objetivos deste trabalho foram avaliar a presença dos BT e GPD nas MHCCR, determinar sua importância prognóstica e sua relação com outros fatores patológicos conhecidos. MÉTODOS: Foram avaliados, retrospectivamente, 229 pacientes submetidos a ressecção de MHCCR com intuito curativo, entre janeiro de 2004 e junho de 2014. Nos espécimes cirúrgicos das MHCCR ressecadas, foi realizada análise anatomopatológica através de lâminas coradas em hematoxilina e eosina (HE), para avaliação dos BT, GPD, infiltrado inflamatório peritumoral, presença de pseudocápsula e tipo de borda; e por imuno-histoquímica, por meio de anticorpos anti-AE1/AE (BT e GPD), anti-D2-40 (invasão linfática) e anti-CD34 (invasão venosa portal). Também foram analisadas as variáveis clínicas relacionadas ao prognóstico. RESULTADOS: O seguimento médio após a hepatectomia foi de 43 meses. As taxas de sobrevida global e livre de doença em 1, 3 e 5 anos foram de, respectivamente, 94,1%, 66,7% e 45,5% e 53,6%, 31,5% e 29,6%. Os BT estiveram presentes em 61,1% dos pacientes na avaliação pelo AE1/AE3 e em 48,9% pelo HE, enquanto os GPD estiveram presentes em 42,8% dos pacientes na avaliação pelo AE1/AE3 e em 49,3% pelo HE. Na análise univariada, os BT e os GPD grau 3 ( > 9 GPD) mostraram significância prognóstica, tanto na avaliação pelo AE1/AE3 quanto pelo HE. Na análise múltipla, os fatores independentes para sobrevida global foram: a presença de GPD grau 3 (pelo HE), presença de pseudocápsula, invasão venosa portal e presença de 4 ou mais nódulos. Os fatores independentes para sobrevida livre de doença foram: GPD grau 3 (pelo HE), quimioterapia prévia, presença de 4 ou mais nódulos, infiltrado inflamatório peritumoral ausente ou leve, invasão venosa portal e borda infiltrativa. Os BT e GPD não estiveram associados à recidiva hepática. Os BT foram associados a: GPD, MHCCR sincrônicas, tumores de até 5 cm, ausência de pseudocápsula, borda infiltrativa, e presença de invasão venosa portal. Os GPD foram associados a: BT, infiltrado inflamatório peritumoral ausente ou leve, ausência de pseudocápsula e borda infiltrativa. CONCLUSÕES: Os BT e os GPD são frequentes nas MHCCR e ambos, na análise univariada, são fatores de pior prognóstico na sobrevida global e livre de doença, estando associados à recidiva extra-hepática. A presença de GPD grau 3 avaliada pelo HE é fator prognóstico independente na sobrevida global e livre de doença, sugerindo que este é um importante mecanismo de disseminação tumoral na MHCCR / INTRODUCTION: The only potentially curative treatment for colorectal liver metastases is surgical resection. However, about 60 to 70% of patients will recur, showing the necessity of a better knowledge regarding spread and disease recurrence pathways. Tumor budding and poorly differentiated clusters, markers related to epithelial-mesenchymal transition, are prognostic factors for colorectal cancer, but their presence and significance in colorectal liver metastases is not yet established. The aims of this study were to evaluate the presence of tumor budding and poorly differentiated clusters in colorectal liver metastases, to determine their prognostic value and to relate them to other known pathological factors. METHODS: A total of 229 patients that underwent hepatic resection for colorectal liver metastases between January 2004 and June 2014 were retrospectively evaluated. Resected specimens of colorectal liver metastases were submitted to pathological evaluation by hematoxilin and eosin staining, to analyze tumor budding, poorly differentiated clusters, peritumoral inflammatory infiltrate, presence of tumor pseudocapsule and tumor growth pattern; and by immunohistochemical staining with antibodies anti- AE1/AE3 (tumor budding and poorly differentiated clusters), anti-D2-40 (lymphatic invasion) and anti-CD-34 (portal vein invasion). Clinical variables related to prognosis were also evaluated. RESULTS: Mean follow up after hepatectomy was 43 months. Overall and disease-free survival rates at 1, 3 and 5 years were 94.1%, 66.7% and 45.5% and 53.6%, 31.5% and 29.6%, respectively. Tumor budding was present in 61.1% of patients in the evaluation by AE1/AE3 staining and in 48.9% by hematoxilin and eosin, while poorly differentiated clusters were present in 42.8% of patients by AE1/AE3 staining and 49.3% by hematoxilin and eosin. At univariate analysis, tumor budding and poorly differentiated clusters grade 3 ( > 9 poorly differentiated clusters) by AE1/AE3 and hematoxilin and eosin staining showed prognostic significance. On multiple analysis, independent factors for overall survival were: presence of poorly differentiated clusters grade 3 (by hematoxilin and eosin), presence of tumor pseudocapsule, portal vein invasion and presence of 4 or more nodules. Independent factors for disease-free survival were: poorly differentiated clusters grade 3 (by hematoxilin and eosin), preoperative chemotherapy, portal vein invasion, presence of 4 or more nodules, none/mild peritumoral inflammatory infiltrate and infiltrative tumor border. Tumor budding and poorly differentiated clusters were not related to hepatic recurrence. Tumor budding was associated with: presence of poorly differentiated clusters, synchronous colorectal liver metastases, tumor size up to 5 cm, absence of tumor pseudocapsule, infiltrative tumor border and presence of portal vein invasion. Poorly differentiated clusters were associated with: presence of tumor budding, none/mild peritumoral inflammatory infiltrate, absence of tumor pseudocapsule and infiltrative tumor border. CONCLUSIONS: Tumor budding and poorly differentiated clusters are frequent in colorectal liver metastases and, on univariate analysis, are prognostic factors for overall and disease-free survival, being associated to extrahepatic recurrence. Presence of poorly differentiated clusters grade 3 stained by hematoxilin and eosin is an independent prognostic factor for overall and disease-free survival, suggesting that it is an important spread pathway in colorectal liver metastases