8/18/2019 1-s2.0-S1547412713000856-main

1/8

Surgery for Esophageal CancerGoals of Resection and OptimizingOutcomesNabil Rizk, MD

Determining what defines an adequate esopha-geal resection to optimize long-term outcomes inesophageal cancer is an elusive goal. The primaryreason for this ambiguousness is the almost totallack of good quality prospective randomized sur-gical trials that examine this question adequately.

Most available data are derived from small retro-spective series typically representing single insti-tution series and their treatment biases. A likelycentral reason for this lack of data is that thereare strongly held opinions by clinicians regardingthe oncologic benefits derived from an esophagealresection. For some, surgery is considered pallia-tive only, and this nihilism is reflected in a mini-malist surgical approach; alternatively, someconsider surgical resection to be a dominantcontributor to improved survival in the manage-ment of esophageal cancer, and for these advo-cates, the more radical the procedure, thegreater the perceived benefit. The reality is, how-ever, likely to be somewhere between these 2 ex-tremes; there is likely a minimal standard thatneeds to be met to achieve optimal oncologicbenefit from an operation, beyond which there islikely no additional marginal benefit. A secondand equally important consideration in optimizingoutcomes for surgery of the esophagus is the

balance between increased short-term complica-tions and oncologic outcome with increasinglyradical surgeries.

The intent of this article is to identify the goals of an appropriate esophagectomy for cancer, essen-tially defining the targets that should be achieved

from an operation. Clearly, these targets are notmonolithic, but rather should reflect the variabilityof the underlying disease process, includingdifferent tumor stages, tumor location, and tumortype. Furthermore, these targets also need to bebalanced with the different risks of surgery, real-izing that with increasing radicality, there areincreased risks. The 2 aspects of the surgicalapproach that have been associated with long-term outcomes and that will be reviewed in thisarticle are the extent of lymphadenectomy, asmeasured by the number of nodes removed andthe location of the nodes removed, and the abilityto achieve disease free surgical margins, termedR-0 resections.

EXTENT OF LYMPHADENECTOMY

The extent of the lymphadenectomy needed in anesophagectomy is controversial. The argumentprimarily centers on the number of lymph node

Disclosures: The author has nothing to disclose.Department of Surgery, Thoracic Service, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, C 883,New York, NY 10065, USAE-mail address: rizkn@mskcc.org

KEYWORDS

Radial margins Proximal margins Distal margins Lymphadenectomy Esophageal cancer

KEY POINTS

The extent of lymphadenectomy should vary based on the risk of nodal metastases. Achieving an R-0 resection is an important component of an esophagectomy. Preoperative chemoradiation improves the likelihood of achieving an R-0 resection, whereas pre-operative chemotherapy alone does not.

Thorac Surg Clin 23 (2013) 491–498http://dx.doi.org/10.1016/j.thorsurg.2013.07.0091547-4127/13/$ – see front matter 2013 Elsevier Inc. All rights reserved. t

h o r a c i c . t h e c l i n i c s . c o m

mailto:rizkn@mskcc.orghttp://dx.doi.org/10.1016/j.thorsurg.2013.07.009http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://thoracic.theclinics.com/http://dx.doi.org/10.1016/j.thorsurg.2013.07.009mailto:rizkn@mskcc.org

8/18/2019 1-s2.0-S1547412713000856-main

2/8

“fields” that need to be resected, as well as theminimum number of nodes that need to beremoved. In fact, unlike most other malignancies,there is still the belief by some that, in esophagealcancer, beyond a complete primary tumor resec-tion, a more extensive lymphadenectomy can

result in improved survival. Despite this ongoingdebate, the reality is that in the United States, thiscontroversy is reserved primarily to the literature,because as in other malignancies such as gastriccancer or lung cancer, surgeons usually performa minimal lymphadenectomy. This minimalist lym-phadenectomy was documented prospectively inthe Z0060 trial whereby the median number of no-des removed was 11, and only 36% of patients had15 or more nodes removed. 1 Likewise, the retro-spective Worldwide Esophageal Cancer Collabo-ration Data (WECC) study showed that only 41%of patients had 15 or more nodes removed. 2 Thisvariability in extent of lymphadenectomy is perva-sive even in the published literature, creating asituation whereby data regarding issues such aslocation of nodal metastases and likelihood of developing nodal metastases are not very reliable.

Likelihood of Nodal Metastases

The likelihood of developing nodal metastases isdirectly correlated with the depth of tumor inva-

sion, and depth of invasion should therefore havean impact on the aggressiveness of the lymphnode dissection. T1a tumors and superficial T1btumors rarely metastasize to lymph nodes. 3 Thesedata would argue then that a minimal to no lym-phadenectomy should be needed in these pa-tients, and that a procedure such as mucosalresection or transhiatal esophagectomy shouldbe adequate treatment. Conversely, deeper sub-mucosal tumors more frequently metastasize tolymph nodes, with some series showing a greater

than 40% incidence of nodal disease.3,4

This highlikelihood of disease mandates a much moreaggressive lymphadenectomy in these patients.Last, adequately resected T2 tumors are 60% to80% likely 5,6 to contain nodal metastases,whereas greater than 80% of T3 tumors will haveevidence of nodal disease. 5,6 There also seem tobe some differences in the propensity to developnodal metastases between squamous cell carci-nomas and adenocarcinomas. Stein and co-workers 7 show that the likelihood to have nodalmetastases was greater in mucosal squamouscell cancers compared with adenocarcinoma(7% vs 0%), as well as in submucosal tumors(36% vs 21%, respectively). This difference mighthave implications on tumor-specific success of local resection treatments in early stage cancers.

Location of Nodal Metastases

The most recent American Joint Committee onCancer staging system categorizes all lymph no-des extending from the peri-esophageal cervicalnodes down to the celiac axis as within the

“zone” of regional disease.8

Although manystudies have shown that primary tumor locationpredisposes toward certain distribution patternsof nodal metastases, in reality there is sufficientvariability in the potential sites on nodal spreadthat selecting nodal basins to resect based onthe site of the primary tumor will invariably missdisease in some patients. The relevant questionregarding this variability is whether a selectivelymphadenectomy based on the site of theprimary tumor is appropriate, or whether thevariability of lymphatic spread needs to be ad-dressed by removing all possible sites of disease.Proponents of selective lymphadenectomy pointto data supporting the strong association be-tween site of tumor origin and likely nodal drain-age basins; in patients with disease beyondthese usual sites, the assumption is that this indi-cates a burden of nodal disease representative of advanced stage disease. 9 Selective drainage im-plies that tumors in the upper esophagus tend tospread superiorly to the upper mediastinal andcervical regions first and are best managed by

resection of these nodes, whereas lower esopha-geal and gastroesophageal tumors spreadprimarily to the upper abdominal and lowerpara-esophageal nodes, an d mid-esophageal tu-mors spread bidirectionally. 10 Using this selectiv eapproach, for instance, Feith and colleagues 9

have shown that Siewert I tumors that arise belowthe level of the carina metastasize primarily tolower para-esophageal and upper abdominal no-des and rarely above the carina. Similarly, thestudy of Schro ¨ der and colleagues 11 of patternsof nodal metastases in Siewert I tumors showedthat, when present, nodal metastases alwayshave an intra-abdominal nodal component, with25% having additional lower mediastinal disease,and only 10% of patients with nodal disease tothe level of the carina. Likewise, others haveshown that Siewert II and III tumors even morerarely metastasize to intrathoracic nodes, 12 andthat therefore the focus of a lymphadenectomyin these tumors should be primarily on the intra-abdominal compartment, with limited inclusionof the lower mediastinal nodes. 13 One inconsis-

tency of many practitioners of a selective lympha-denectomy, however, is that they often do notresect cervical and upper mediastinal nodes inmid-esophageal tumors (especially squamouscell cancers) despite the evidence supporting its

Rizk492

8/18/2019 1-s2.0-S1547412713000856-main

3/8

benefits 14 ; this raises the criticism that many whoespouse selective lymphadenectomy in fact arenot doing so consistently in mid-esophagealtumors.

What the selective approach to lymphadenecto-mies fails to address is the presence of unex-

pected drainage patterns and skip metastases.Several studies in which a thorough lymphadenec-tomy has been performed have shown a high inci-dence of skip metastases. Skip metastases havebeen well documented to occur by both in adeno-carcinoma 5,6,15 with an incidence of about 30%,and up to 76% in squamous cell carcinoma. 10

Because of this variability as well as the belief inthe benefits of a therapeutic lymphadenectomy,there are some who argue for routine 3-field lym-phadenectomy for either all tumors 5,16 or for asubset of patients thought to be at higher risk forskip metastases, such as Siewert I adenocarci-noma with a moderate nodal burden. 6

Describing the Extent of Lymphadenectomy

Lymphadenectomies during esophagectomies aremost commonly divided into 3 fields. The first fieldis intra-abdominal and includes nodes on thelesser curvature of the stomach, nodes betweenthe pancreas and the crura, and nodes along thehepatic, splenic, common hepatic, and left gastric

arteries. These nodes are the most commonlyinvolved lym ph nodes in gastroesophageal junc-tion tumors, 12,13 yet they tend to be the mostfrequently undersampled nodes by thoracic sur-geons. The second field of lymph nodes is intra-thoracic. These nodes are variously described asextending from either the carina down to the hiatus(most common description) or also including allsuperior mediastinal nodes up to the thoracic inletincluding nodes along the right and left recurrentnerves. 5 The nodes from the second field are

less commonly involved with disease than the firstfield in adenocarcinomas. 11,13 Although resectionof the second field typically is performed using aright thoracotomy, the ability to resect the lowerpara-esophageal nodes via a thoracotomy issomewhat compromised due to difficulties in visu-alizing the region. A more effective retrievalmethod of these nodes is either with a minimallyinvasive approach or by a laparotomy and aradical transhiatal approach. 17 The classicallydescribed third field in an esophageal lymphade-nectomy is resected via a cervical incision and in-cludes the removal of nodes along both therecurrent nerves and the deep cervical nodeslateral and posterior to the common carotidsheath. 18 Some consider the superior mediastinalnodes to also be part of the third field. 6

Another means of defining the extent of a lym-phadenectomy is by quantifying the number of no-des that are removed. This descriptor adds adimension of quality control to a lymphadenec-tomy because the number of nodes removed canreveal the aggressiveness of the lymphadenec-

tomy, whereas the lymph node field descriptiononly describes the location of the nodes removed.On the other hand, using the number of nodesremoved as a quality standard has its own limita-tions, including the fact that there is no controlover which nodes are removed. Furthermore,quantifying the number of nodes removed can besubjective, partly dependent on how involved sur-geons are at identifying the nodes for the patholo-gists, and partly dependent on the level of effort thepathologists make at counting thenodes. Themostdependable description of a lymphadenectomy isto describe which fields were dissected as wellas to provide the number of nodes removed.

Risks of Lymphadenectomy

The increased risks of a more extensive lymphade-nectomy have been shown in several studies andare likely attributable to the additional exposureneeded to perform it, the anatomic proximity of surrounding structures (thoracic duct, recurrentnerves, airway), and the additional operative time. A more thorough upper abdominal lymphadenec-tomy (D1 vs D2) has been shown in some random-ized gastrectomy trials to be associated withincreased morbidity and mortality. 19 Likewise, a2-field lymphadenectomy generally requires anintrathoracic component to the procedure; if this isperformed with a thoracotomy, there has clearlybeen shown to be an association with increasedmorbidity and mortality. 20 Furthermore, in perform-ing a complete 2-field lymphadenectomy, the po-tential for a thoracic duct injury is greater thanwhen a more limited lymphadenectomy is per-

formed. 20 The added risks of a thoracotomy mightbemitigated,however,by using a minimally invasiveapproach. Last, some studies have shown dramat-ically higher complication rates when removing thethird fieldofnodes,witha significantlyincreasedpo-tential for unilateral and bilateral recurrent nerveinjury; one randomized study had a greater than50% tracheostomy incidence. 14 This reason andthe preponderance of adenocarcinomas seen inthe West are the most commonly cited reasons athird field lymphadenectomy is rarely performed.

Evidence Supporting a Lymphadenectomy

The evidence that the type and extent of lympha-denectomy is correlated with outcome is likelyconfounded by 2 important variables. The first is

Surgery for Esophageal Cancer 493

8/18/2019 1-s2.0-S1547412713000856-main

4/8

stage migration. Most of the data that support acorrelation between the extent of a lymphadenec-tomy and outcome are retrospective, and there-fore, cannot differentiate between stage migrationor survival benefit. The second confounder is thefact that the various retrospective studies have

varying distributions of pathologic stages, makingcomparisons of results difficult given the varyinglikelihood of metastatic nodal disease amongdifferent stages. Last, some retrospective studiesalso include patients who received preoperativetherapy; in these patients it is difficult to accountfor nodal down-staging.

The most convincing evidence supporting theadded benefit of extending a lymphadenectomyduring an esophagectomy comes from theHulscher trial. 17 In this well-controlled prospectiverandomized trial comparing an en-blocesophagec-tomy, which included a complete lower mediastinallymphadenectomy and a right paratracheal lym-phadenectomy to a radical transhiatal esophagec-tomy, long-term follow-up showed that including amore thorough intrathoracic lymphadenectomyimproved survival in patients with Siewert I tumorswith 1 to 8 nodes involved with disease. 21 Eventhough this result was derived from an unplannedsubset analysis, what makes the data morecompelling is that they are consistent with theexpectation that a lymphadenectomy should only

be helpful if the patient is at risk for nodal diseaseor if the potential exists that the nodes removedrepresent all the possible disease sites (this wouldbe less likely the more nodes are involved). Thisresult is also consistent with the WECC retrospec-tive study on the correlation between extent of lym-phadenectomy and survival, as well as in the studyby Schwarz and Smith. 22 The WECC study, unlikeprevious studies that supported a single minimallymphadenectomy standard to optimize sur-vival, 23,24 contained a sufficient number of patients

to identify stage-dependent minimal lymphadenec-tomy standards. 2 Thus, patients at minimal to norisk of lymph nodal metastases (T1 tumors) derivedno benefit from a more extensive lymphadenec-tomy,nor did patients withsignificant nodal disease( 7 nodes), whereas patients with a moderate riskfor nodal metastases showed a graduated benefitfrom a more extensive lymphadenectomy as therisk and presence for nodal disease increased.These retrospective findings, although subject tothe previously listed pitfalls, nevertheless seem toconfirm the findings of the randomized trial.

Recommendations

Based on theavailable data, a reasonableapproachin deciding the extent of a lymphadenectomy for

esophageal cancer should incorporate the clinicalstage and location of the disease, and the potentialfor nodal metastases, with the caveat that anybenefit from a lymphadenectomy is due to stagemigration, or from an actual survival benefit, orfrom both. Patients at very low risk (T1a, superficial

T1b) likely do not need a lymphadenectomy,implying that a mucosal resection or a transhiatalesophagectomy would be an adequate surgicalapproach. Patients at moderate risk for nodal dis-ease (deep T1b, T2)need a more extensive lympha-denectomy (15–20 nodes) and patients at highestrisk (T3) need the most extensive lymphadenec-tomy (>30 nodes). With regard to which fieldsneed to be resected, clearly the first and modifiedsecond (infra-azygous) are the most relevant formost adenocarcinomas and lower esophagealsquamous cell cancers, and thesecondandpoten-tially the third fields are the most relevant for mid-esophageal squamous cell carcinomas. Morecontroversial are the removal of superior medias-tinal nodes and possibly cervical nodes forSiewert I lesions as advocated by some, andlimiting the lower mediastinal lymphadenectomy inSiewert III tumors. These 2 factors (number andlocation of nodes) should then dictate which surgi-cal approaches should be considered.

RESECTION MARGINDefinition

Resection margin (R-resection) refers to whetherradial, proximal, or distal margins are grosslyinvolved with disease (R-2), are completely unin-volved (R-0), or either have direct microscopicinvolvement at the margins as defined by the Col-lege of American Pathologists (R-1, College of American Pathologists criteria) or, as defined bythe Royal College of Pathologists (RCP), have mar-gins measured at less than 1 mm from the cut sur-

face (R-1, RCP criteria).25

These 2 definitions of anR-1 resection, while on the surface seeming to beinsignificant, in fact can be associated with signif-icant consequences if decisions regarding adju-vant radiation are based on a pathologic findingof an R-1 resection. In esophagectomies, there isalso a separate distinction made between radialmargins (the margins extending laterally from theesophageal wall and tumor), and the proximal(esophageal) and distal (gastric) margins. Moststudies that refer to R-resection rates focus pri-marily on the radial margins. Proximal and distalmargins are much less often involved with disease,although several specific issues deserve mention,including the extent of gastric resection neededin Siewert II and Siewert III cancers, the extentof proximal margin needed in squamous cell

Rizk494

8/18/2019 1-s2.0-S1547412713000856-main

5/8

cancers, and the relevance of dysplasia and meta-plasia at the proximal margin. Last, there is animportant difficult management question that oc-casionally arises at the time of the operation or inthe final pathology report, namely, what to doabout a positive intraoperative margin.

Association with Outcome

Various studies have documented an associationbetween R-resection and survival. A significantlimitation in most of these data, however, is the in-clusion of various proportions of patients at no riskof involved radial margins (anyone

8/18/2019 1-s2.0-S1547412713000856-main

6/8

compromise of the extent of the proximalmargin. 36 Conversely, and especially in light of the new American Joint Committee on Cancercate gorization of these tumors as esophageal can-cers, 8 others think that these tumors should bemanaged with whichever approach is capable of

best achieving an R-0 resection, using either agastrectomy with a jejeunal interposition, anesophagogastrectomy using the residual stomachas a conduit, or a gastrectomy and esophagec-tomy using the colon as a conduit. 37–40

Proximal Esophageal Margins and SquamousCell Cancer

Another area of controversy regarding margins isin esophageal squamous cell cancers. The con-cern in these cancers is the presence of either

multifocal disease or proximal extension of sub-mucosal disease beyond the main tumor mass,with the potential for persistent disease in theproxima l esophageal remnant. Lam and co-workers 41 showed that the incidence of intramuralspread beyond the primary tumor can occur bothproximally and distally in up to 72% of cases,and up to 9 cm in distance. Tsutsui and col-leagues 42 showed a similarly high incidence (upto 55%) and correlated the distance of spread tothe depth of invasion of the primary tumor. Thisconsideration prompts some to routinely recom-mend a complete esophagectomy in patientswith esophageal squamous cell cancer, regardlessof the proximal extent of the known tumor mass.

Dysplasia and Metaplasia at the Proximal Margin

One concern in Barrett’s associated esophagealadenocarcinomas is the potential presence of dysplasia and/or metaplasia at the proximalmargin. Clearly one goal of the resection needsto be resection of the esophagus proximal toinvolved sites; occasionally, however, technicalconcerns regarding conduit length and viabilitycan limit the proximal extent of esophagus thatcan be resected. The data regarding the neces-sity of avoiding dysplastic or metaplastic prox-imal margins are virtually inexistent. However,with the availability of successful endoscopictreatments of dysplasia and metaplasia, it wouldseem that this concern should become lessprominent.

A Positive Intraoperative Frozen SectionMargin

A well-planned and executed operation shouldalmost always be able to achieve proximal anddistal margins that are uninvolved with disease

(R-0 resection). Occasionally, however, one willencounter a situation intraoperatively of a “posi-tive” microscopic margin (R-1 resection) with noeasy recourse without significantly expanding thescope and risks of the operation. In these in-stances, the obvious question is, is it worth the

added risk to achieve a negative intraoperativemargin? The assumption in such a question needsto be that the operation was well planned andexecuted, and therefore, the involved marginswere completely unexpected based on the clinicalassessment of the extent and type of disease.Under these circumstances, the presence of aninvolved margin is frequently an indicator of moreadvanced underlying disease, and as with gastriccancers, the involved microscopic marginfrequently becomes irrelevant prognostically. 43

Given the unavailability of detailed staging dataduring an operation, reconsideration on how tomanage the margins can be made after the finalpathologic information is available and a more ac-curate prognosis can be made.

Recommendations

Given the likely association between margins andsurvival, achieving an R-0 resection should bea major component of attempting to curativelytreat patients with esophageal cancers. The radialmargin is the most commonly involved margin inlocally advanced (T3, T4) tumors. Minimizing thisrisk can be done either with the addition of preop-erative chemoradiation or through a more ag-gressive radial resection, including an en-blocresection if necessary. Proximal and distal marginsare much less often involved, but there are specificissues that need to be taken into consideration toimprove long-term survival. In tumors involving thegastroesophageal junction with significant gastriccardia involvement, a gastrectomy should beconsidered to achieve an R-0 resection, with the

caveat that proximal margins are not compro-mised. Similarly, a total esophagectomy shouldbe considered in squamous cell cancers becauseof the high potential for submucosal extension of disease. Last, in situations of appropriate surgicalplanning and execution, and an R-1 resection onfrozen section, a reasonable approach would beto review the final pathology report first before ex-tending the operation significantly; often in thesesituations, the long-term prognosis is sufficientlypoor that a more aggressive local treatment isunwarranted.

REFERENCES

1. Veeramachaneni NK, Zoole JB, Decker PA, et al,American College of Surgeons Oncology Group

Rizk496

http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1

8/18/2019 1-s2.0-S1547412713000856-main

7/8

Z0060 Trial. Lymph node analysis in esophagealresection: American College of Surgeons OncologyGroup Z0060 trial. Ann Thorac Surg 2008;86(2):418–21 [discussion: 421] .

2. Rizk NP, Ishwaran H, Rice TW, et al. Optimum lym-phadenectomy for esophageal cancer. Ann Surg

2010;251(1):46–50 .3. Ancona E, Rampado S, Cassaro M, et al. Prediction

of lymph node status in superficial esophageal car-cinoma. Ann Surg Oncol 2008;15(11):3278–88 .

4. Bollschweiler E, Baldus SE, Schro ¨der W, et al. Highrate of lymph-node metastasis in submucosalesophageal squamous-cell carcinomas and adeno-carcinomas. Endoscopy 2006;38(2):149–56 .

5. Lerut T, Nafteux P, Moons J, et al. Three-field lym-phadenectomy for carcinoma of the esophagusand gastroesophageal junction in 174 R0 resec-tions: impact on staging, disease-free survival, andoutcome: a plea for adaptation of TNM classificationin upper-half esophageal carcinoma. Ann Surg2004;240(6):962–72 [discussion: 972–4] .

6. Altorki N, Kent M, Ferrara C, et al. Three-field lymphnode dissection for squamous cell and adenocarci-noma of the esophagus. Ann Surg 2002;236(2):177–83 .

7. Stein HJ, Feith M, Bruecher BL, et al. Early esopha-geal cancer: pattern of lymphatic spread and prog-nostic factors for long-term survival after surgicalresection. Ann Surg 2005;242(4):566–73 [discus-

sion: 573–5] .8. Edge SB, Byrd DR, Compton CC, et al, editors.

AJCC cancer staging Manual. 7th edition. Springer;2010 .

9. Feith M, Stein HJ, Siewert JR. Pattern of lymphaticspread of Barrett’s cancer. World J Surg 2003;27(9):1052–7 .

10. Chen J, Liu S, Pan J, et al. The pattern and preva-lence of lymphatic spread in thoracic oesophagealsquamous cell carcinoma. Eur J Cardiothorac Surg2009;36(3):480–6 .

11. Schrö der W, Mö nig SP, Baldus SE, et al. Fre-quency of nodal metastases to the upper medias-tinum in Barrett’s cancer. Ann Surg Oncol 2002;9(8):807–11 .

12. Fox MP, van Berkel V. Management of gastroesoph-ageal junction tumors. Surg Clin North Am 2012;92(5):1199–212 .

13. Rüdiger Siewert J, Feith M, Werner M, et al. Adeno-carcinoma of the esophagogastric junction: resultsof surgical therapy based on anatomical/topo-graphic classification in 1,002 consecutive patients.Ann Surg 2000;232(3):353–61 .

14. Nishihira T, Hirayama K, Mori S. A prospective ran-domized trial of extended cervical and superiormediastinal lymphadenectomy for carcinoma ofthe thoracic esophagus. Am J Surg 1998;175(1):47–51 .

15. Hosch SB, Stoecklein NH, Pichlmeier U, et al.Esophageal cancer: the mode of lymphatic tumorcell spread and its prognostic significance. J ClinOncol 2001;19(7):1970–5 .

16. Igaki H, Tachimori Y, Kato H. Improved survival forpatients with upper and/or middle mediastinal lymph

node metastasis of squamous cell carcinoma of thelower thoracic esophagus treated with 3-fielddissection. Ann Surg 2004;239(4):483–90 .

17. Hulscher JB, van Sandick JW, de Boer AG, et al.Extended transthoracic resection compared withlimited transhiatal resection for adenocarcinoma ofthe esophagus. N Engl J Med 2002;347(21):1662–9 .

18. Cense HA, van Eijck CH, Tilanus HW. New insightsin the lymphatic spread of oesophageal cancerand its implications for the extent of surgical resec-tion. Best Pract Res Clin Gastroenterol 2006;20(5):893–906 .

19. Bonenkamp JJ, Songun I, Hermans J, et al. Rando-mised comparison of morbidity after D1 and D2dissection for gastric cancer in 996 Dutch patients.Lancet 1995;345:745–8 .

20. Hulscher JB, Tijssen JG, Obertop H, et al. Transtho-racic versus transhiatal resection for carcinoma ofthe esophagus: a meta-analysis. Ann Thorac Surg2001;72(1):306–13 .

21. Omloo JM, Lagarde SM, Hulscher JB, et al.Extended transthoracic resection compared withlimited transhiatal resection for adenocarcinoma of

the mid/distal esophagus: five-year survival of arandomized clinical trial. Ann Surg 2007;246(6):992–1000 [discussion: 1000–1] .

22. Schwarz RE, Smith DD. Clinical impact of lymphade-nectomy extent in resectable esophageal cancer.J Gastrointest Surg 2007;11(11):1384–93 [discus-sion: 1393–4] .

23. Rizk NP, Venkatraman E, Park B, et al. The prog-nostic importance of the number of involved nodesin esophageal cancer: implications for revisionsof the American Joint Committee on Cancer StagingSystem. J Thorac Cardiovasc Surg 2006;132:1374–81 .

24. Peyre CG, Hagen JA, DeMeester SR, et al. The num-ber of lymph nodes removed predicts survival inesophageal cancer: an International Study on theImpact of Extent of Surgical Resection. Ann Surg2008;248:190–7 .

25. Deeter M, Dorer R, Kuppusamy MK, et al. Assess-ment of criteria and clinical significance of circum-ferential resection margins in esophageal cancer.Arch Surg 2009;144(7):618–24 .

26. Verhage RJ, Zandvoort HJ, ten Kate FJ, et al. How to

define a positive circumferential resection margin inT3 adenocarcinoma of the esophagus. Am J SurgPathol 2011;35(6):919–26 .

27. Dexter SP, Sue-Ling H, McMahon MJ, et al.Circumferential resection margin involvement: an

Surgery for Esophageal Cancer 497

http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref2http://refhub.elsevier.com/S1547-4127(13)00085-6/sref2http://refhub.elsevier.com/S1547-4127(13)00085-6/sref2http://refhub.elsevier.com/S1547-4127(13)00085-6/sref3http://refhub.elsevier.com/S1547-4127(13)00085-6/sref3http://refhub.elsevier.com/S1547-4127(13)00085-6/sref3http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref8http://refhub.elsevier.com/S1547-4127(13)00085-6/sref8http://refhub.elsevier.com/S1547-4127(13)00085-6/sref8http://refhub.elsevier.com/S1547-4127(13)00085-6/sref9http://refhub.elsevier.com/S1547-4127(13)00085-6/sref9http://refhub.elsevier.com/S1547-4127(13)00085-6/sref9http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref12http://refhub.elsevier.com/S1547-4127(13)00085-6/sref12http://refhub.elsevier.com/S1547-4127(13)00085-6/sref12http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref26http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref25http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref24http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref23http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref22http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref21http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref20http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref19http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref18http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref17http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref16http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref15http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref14http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref13http://refhub.elsevier.com/S1547-4127(13)00085-6/sref12http://refhub.elsevier.com/S1547-4127(13)00085-6/sref12http://refhub.elsevier.com/S1547-4127(13)00085-6/sref12http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref11http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref10http://refhub.elsevier.com/S1547-4127(13)00085-6/sref9http://refhub.elsevier.com/S1547-4127(13)00085-6/sref9http://refhub.elsevier.com/S1547-4127(13)00085-6/sref9http://refhub.elsevier.com/S1547-4127(13)00085-6/sref8http://refhub.elsevier.com/S1547-4127(13)00085-6/sref8http://refhub.elsevier.com/S1547-4127(13)00085-6/sref8http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref7http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref6http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref5http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref4http://refhub.elsevier.com/S1547-4127(13)00085-6/sref3http://refhub.elsevier.com/S1547-4127(13)00085-6/sref3http://refhub.elsevier.com/S1547-4127(13)00085-6/sref3http://refhub.elsevier.com/S1547-4127(13)00085-6/sref2http://refhub.elsevier.com/S1547-4127(13)00085-6/sref2http://refhub.elsevier.com/S1547-4127(13)00085-6/sref2http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1http://refhub.elsevier.com/S1547-4127(13)00085-6/sref1

8/18/2019 1-s2.0-S1547412713000856-main

8/8

independent predictor of survival following surgeryfor oesophageal cancer. Gut 2001;48(5):667–70 .

28. Pultrum BB, Honing J, Smit JK, et al. A criticalappraisal of circumferential resection margins inesophageal carcinoma. Ann Surg Oncol 2010;17(3):812–20 .

29. Rao VS, Yeung MM, Cooke J, et al. Comparison ofcircumferential resection margin clearance criteriawith survival after surgery for cancer of esophagus.J Surg Oncol 2012;105(8):745–9 .

30. Khan OA, Fitzgerald JJ, Soomro I, et al. Prognosticsignificance of circumferential resection margininvolvement following oesophagectomy for cancer.Br J Cancer 2003;88(10):1549–52 .

31. Suttie SA, Nanthakumaran S, Mofidi R, et al. Theimpact of operative approach for oesophageal can-cer on outcome: the transhiatal approach may influ-ence circumferential margin involvement. Eur J SurgOncol 2012;38(2):157–65 .

32. Williams VA, Peters JH. Adenocarcinoma of thegastroesophageal junction: benefits of an extendedlymphadenectomy. Surg Oncol Clin N Am 2006;15(4):765–80 .

33. van Hagen P, Hulshof MC, van Lanschot JJ, et al.Preoperative chemoradiotherapy for esophageal orjunctional cancer. N Engl J Med 2012;366:2074–84 .

34. Kelsen DP, Ginsberg R, Pajak TF, et al. Chemo-therapy followed by surgery compared with surgeryalone for localized esophageal cancer. N Engl J

Med 1998;339:1979–84 .35. Kim KT, Jeong O, Jung MR, et al. Outcomes of

abdominal total gastrectomy for Type II and II

gastroesophageal junction tumors: single center’sexperience in Korea. J Gastric Cancer 2012;12(1):36–42 .

36. Papachristou DN, Agnanti N, D’Agostino H, et al.Histologically positive esophageal margin in the sur-gical treatment of gastric cancer. Am J Surg 1980;

139(5):711–3 .37. Johansson J, Djerf P, Oberg S, et al. Two different

surgical approaches in the treatment of adenocarci-noma at the gastroesophageal junction. World JSurg 2008;32(6):1013–20 .

38. Avella D, Garcia L, Hartman B, et al. Esophagealextension encountered during transhiatal resectionof gastric or gastroesophageal tumors: attaining anegative margin. J Gastrointest Surg 2009;13(2):368–73 .

39. Mattioli S, Di Simone MP, Ferruzzi L, et al. Surgicaltherapy for adenocarcinoma of the cardia: modal-ities of recurrence and extension of resection. DisEsophagus 2001;14(2):104–9 .

40. Ito H, Clancy TE, Osteen RT, et al. Adenocarcinomaof the gastric cardia: what is the optimal surgicalapproach? J Am Coll Surg 2004;199(6):880–6 .

41. Lam KY, Ma LT, Wong J. Measurement of extent ofspread of oesophageal squamous carcinoma by se-rial sectioning. J Clin Pathol 1996;49(2):124–9 .

42. Tsutsui S, Kuwano H, Watanabe M, et al. Resectionmargin for squamous cell carcinoma of the esoph-agus. Ann Surg 1995;222(2):193–202 .

43. Kim SH, Karpeh MS, Klimstra DS, et al. Effect ofmicroscopic resection line disease on gastric can-cer survival. J Gastrointest Surg 1999;3:24–33 .

Rizk498

http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref33http://refhub.elsevier.com/S1547-4127(13)00085-6/sref33http://refhub.elsevier.com/S1547-4127(13)00085-6/sref33http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref40http://refhub.elsevier.com/S1547-4127(13)00085-6/sref40http://refhub.elsevier.com/S1547-4127(13)00085-6/sref40http://refhub.elsevier.com/S1547-4127(13)00085-6/sref41http://refhub.elsevier.com/S1547-4127(13)00085-6/sref41http://refhub.elsevier.com/S1547-4127(13)00085-6/sref41http://refhub.elsevier.com/S1547-4127(13)00085-6/sref42http://refhub.elsevier.com/S1547-4127(13)00085-6/sref42http://refhub.elsevier.com/S1547-4127(13)00085-6/sref42http://refhub.elsevier.com/S1547-4127(13)00085-6/sref43http://refhub.elsevier.com/S1547-4127(13)00085-6/sref43http://refhub.elsevier.com/S1547-4127(13)00085-6/sref43http://refhub.elsevier.com/S1547-4127(13)00085-6/sref43http://refhub.elsevier.com/S1547-4127(13)00085-6/sref43http://refhub.elsevier.com/S1547-4127(13)00085-6/sref43http://refhub.elsevier.com/S1547-4127(13)00085-6/sref42http://refhub.elsevier.com/S1547-4127(13)00085-6/sref42http://refhub.elsevier.com/S1547-4127(13)00085-6/sref42http://refhub.elsevier.com/S1547-4127(13)00085-6/sref41http://refhub.elsevier.com/S1547-4127(13)00085-6/sref41http://refhub.elsevier.com/S1547-4127(13)00085-6/sref41http://refhub.elsevier.com/S1547-4127(13)00085-6/sref40http://refhub.elsevier.com/S1547-4127(13)00085-6/sref40http://refhub.elsevier.com/S1547-4127(13)00085-6/sref40http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref39http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref38http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref37http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref36http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref35http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref34http://refhub.elsevier.com/S1547-4127(13)00085-6/sref33http://refhub.elsevier.com/S1547-4127(13)00085-6/sref33http://refhub.elsevier.com/S1547-4127(13)00085-6/sref33http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref32http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref31http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref30http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref29http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref28http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27http://refhub.elsevier.com/S1547-4127(13)00085-6/sref27