2010 SSAT PLENARY PRESENTATION

Repair of Symptomatic Giant Paraesophageal Herniasin Elderly (>70 Years) Patients Results in ImprovedQuality of Life

Brian E. Louie & Maurice Blitz & Alexander S. Farivar &Jeraldine Orlina & Ralph W. Aye

Received: 3 May 2010 /Accepted: 9 August 2010 /Published online: 19 January 2011# 2011 The Society for Surgery of the Alimentary Tract

AbstractIntroduction Giant paraesophageal hernias (PEH) involve herniation ofstomach and/or other viscera into the mediastinum.These are usually symptomatic and commonly occur in the elderly. The benefits and risks of operating on elderly patientswith giant PEH have not been clearly elucidated.Materials and Methods We performed a retrospective chart review of consecutive patients aged 70 or greater with giantPEHs undergoing repair.Quality of life data were gathered using QOLRAD, GERD-HRQL and adysphagia severity score.Results Fifty-eight patients (34 females), median 78 years old, presented for repair. Nine patients presented urgently. There wasno 30-daymortality. Major morbidity was 15.5%. At mean follow-up of 1.3 years, 81%were symptom free compared to baseline(p

junction to the diaphragmatic impressionswith a paraesophagealcomponent and/or having a paraesophageal configurationdefined as type IIIV.5 Patients admitted to the hospital withsymptoms of incarceration or obstruction that necessitated earlyendoscopy, nasogastric decompression, and repair during thesame admission were included in this analysis. We excluded allpatients with sliding hiatal hernias as well as those requiringemergent operative intervention for strangulated PEH orincarceration that did not respond to decompression.

Preoperative evaluation of each patient included adetailed history and physical examination, an uppergastrointestinal videoesophagogram, upper endoscopy, andhigh-resolution manometry when possible. Wireless pHanalysis was done at the discretion of the attending surgeon.Other imaging such as computed tomography, pulmonaryfunction tests, and gastric emptying tests were obtained asneeded on an individual basis.

Operative Techniques

All procedures were performed by a team including anattending surgeon (R.A. or B.L.) with a senior resident orMIS fellow. The laparoscopic approach is performed in lowlithotomy with five small incisions following principlespreviously described.1 Esophageal lengthening procedureswere not employed in any of our patients. Cruralreconstruction was performed with simple non-absorbable,braided 0 sutures (polyethylene terephthalate coated withpolybutilate, Ethicon-Johnson and Johnson, Cincinnati,OH). Bioabsorbable mesh reinforcement was used liberallyafter the trial by Oelschlager was published.6

An open approach was used sparingly. A transabdominalapproach was performed when the PEH was concomitantlyrepaired with another intra-abdominal procedure. A transthoracicapproach was used if the abdomen was hostile and inaccessibleto either laparoscopic or transabdominal approaches. Meshreinforcement was not used in the transthoracic repair.

An anti-reflux procedure was routinely performed aftercrural reconstruction. Three procedures were used at thediscretion of the operating surgeon. When a Nissenfundoplication was created, it was performed over a 60-Frbougie and fashioned to 2 cm in length. When a Hill repairwas created, it was done according to the principlesdescribed by Aye.7 A special 43-Fr bougie with an open tipto allow for a water perfused manometry catheter to beadvanced was utilized to perform intraoperative manometry.The last anti-reflux procedure was a hybrid procedurecombining the Nissen fundoplication and the Hill repair.8 Atour center, we have surgeons experienced in both the Nissenand Hill procedures. This hybrid operation was conceivedand evaluated in an institutional review board approved pilotstudy where the Nissen fundoplication is performed over twoHill gastroplasty sutures placed through the collar sling fibers

of the gastrointestinal junction and secured to the pre-aorticfascia in hopes of mitigating axial tension and cephaladdisplacement.

Quality of Life Instruments

Quality of life data were gathered using three diseasespecific instruments: the Quality of Life in Reflux andDyspepsia Questionnaire (QOLRAD), GERD-HRQL, and adysphagia score. These instruments were completed by thepatient at the first office consultation and postoperativelyfor short-term follow-up at 4 to 6 weeks and long-termfollow-up at 6 and 12 months and yearly thereafter.

TheQOLRAD is a validated 25-item questionnaire designedfor self-administration by patients with upper gastrointestinalsymptoms with a maximum score of 7.9 Each item asks thepatients to reflect on the impact of GERD or esophagealproblems over the past week and to rate it on a 7-point scale.A higher score represents an improved quality of life.Although it is a disease specific questionnaire focusing onGERD and dyspepsia, it has been broadly applied acrossmany upper intestinal disorders as an overall QOL instrument.The 25 questions attempt to ascertain GI health in terms ofemotional distress (six items), sleep disturbance (five items),food/drink problems (six items), physical/social functioning(five items), and vitality (three items).

The GERD-HRQL is a disease specific quality of lifeinstrument that has been validated to measure symptomseverity in gastroesophageal reflux.10 It has been used andvalidated to assess response to medications, endoscopicprocedures, and surgery. The self-administered instrumentconsists of ten questions and a separate global satisfactionquestion. Likert-type responses are possible, with 0 representingno symptoms to 5 reflecting incapacitating symptoms andunable to do daily activities. The scores can range from 54 to 0with a lower overall score equating to better quality of life.

To assess the symptom of dysphagia, we used thevalidated dysphagia score as described by Dakkak.11 Thisinstrument was designed to be used with a standardizedmeal eaten within 7 days of completing the questionnaireand combined with a blinded observer documenting theactual food ingestion. Patients are asked about the ease ofingesting certain textures of foods and defined amounts. Ascore of 0 reflects that no food was ingested, whereas amaximal score of 45 represents the ability to ingest theentire meal. To simplify the use of the instrument, patientswere asked about their ability to ingest each of the ninefoods and points are awarded according to the standardizedweighting system. If a patient related that no difficulty wasencountered full points were awarded. Conversely, if thepatient admitted difficulty in ingesting a certain food, nopoints were awarded. Half of the points were awarded ifmoderate difficulty was encountered.

390 J Gastrointest Surg (2011) 15:389396

Data and Statistical Analysis

Demographic, operative, clinical, and quality of life datawere collected from the clinic chart and hospital medicalrecord. Long-term quality of life analysis was conducted byphone interview by one of the attending surgeons inpatients who were not able to travel to the clinic. Statisticalanalysis was performed using SPSS 18. Continuousvariables were analyzed using Students t test. Categoricalvariables were analyzed using chi-squared. Symptoms wereanalyzed by McNemars test. The institutional review boardapproved this study.

Results

A total of 58 patients were assessed and underwent surgicalrepair of a symptomatic giant PEH. Median age was 78 years.Their baseline demographics and hernia characteristics areoutlined in Table 1. The most common PEH was a type III(78%), with an average size of 10 cm as measured fromdiaphragmatic hiatus to the top of the gastric fundus onimaging. Nine patients presented urgently with symptoms ofincarceration. There was no 30-day mortality. Three patients

died in follow-up: one from lung cancer and two fromnatural causes.

The three different repairs used in this series were evenlyapplied with 18 Nissen fundoplications, 19 Hill procedures,and 20 combined HillNissen repairs. One patient had anAllison repair with PEG. A bioabsorbable mesh was placedin 38% of cases. A laparoscopic repair was attempted in 55patients and successfully completed in 53 with two cases(3.4%) converted to open laparotomy. One conversion wasfor an intraoperative esophageal perforation during a Hillrepair and the other for poor visualization while attemptingto reduce a large complex type IV hernia. Three casesutilized an open incision from the outset. Two were done in thisfashion because a concurrent abdominal procedure was alsoplanned. One was performed via a left thoracotomy because thepatient had significant previous abdominal surgery.

Thirteen patients experienced morbidity (Table 2). Therewere five (8%) minor morbidities including two patientsrequiring mechanical ventilation for less than 24 h. Onerequired intubation overnight for hypercarbia after repair ofa large type IV hernia, and one was re-intubated brieflyafter developing re-expansion pulmonary edema after repairof large type III hernia compressing the left lower lobe.There were nine (16%) major morbidities. Four patientsrequired readmission for dehydration. There were twoesophageal perforations during passage of the bougie forintraoperative manometry. One was repaired laparoscopicallyand the other converted to an open procedure and thenrepaired. Both patients were discharged without furthercomplications or interventions.

At a mean follow-up 1.3 years (6 months5.5 years), 81%of patients were entirely symptom-free compared to baseline(p

57% of patients. When compared to baseline preoperativescores, the QOLRAD improved from 5.0 to 6.1 (p

required re-operation. Mesh was placed in three of therecurrences. Mesh was not placed in one patient at thediscretion of the surgeon and the other patients hadsurgery before mesh was popularized. Despite thepresence of recurrence, the median long-term QOLRADscore was 6.8, GERD-HRQL was 3.5, and the dysphagiascore was 39. These were not statistically different fromthe patients without recurrence.

Discussion

Herniation of the stomach into the chest in elderly patients(>70 years of age) is a dilemma to many physicians. Thesepatients often have associated comorbid disease that delaysreferral to surgical specialists because of concerns about theprohibitive risks of surgery. Secondly, the presentingcomplaints such as chest pain or shortness of breath oftendirect the physician to consider a cardiac or pulmonaryetiology for their symptoms. Accordingly, these systems areevaluated. The PEH is often discovered incidentally onimaging. Even when it is determined that the PEH is thecause of their symptoms, both patients and physicians arereluctant to seek surgical consultation for fear of increasedmorbidity and mortality and a perception that surgicalintervention will not resolve their symptoms or improvetheir QOL.

However, in this series as well as others publishedpreviously,1,2 it has been demonstrated that surgical reductionof the hernia and its sac, crural reconstructionwith bioabsorbablemesh, and an anti-reflux repair in an elderly patient populationcan be performed safely, with a very low mortality rate andacceptable morbidity. Even though no mortalities were reportedin this series andmajor morbidity was 16%, it remains importantto carefully evaluate the elderly patient since age greater than 70,BMI greater than 35 kg/m2, and multiple comorbidities havebeen identified as factors that may increase the chances of anadverse outcome.1

The symptoms derived from giant PEHs are oftensecondary to both acid reflux and mechanical factors.5

Although the acid-related symptoms may be partially orless often totally resolved with proton pump inhibitors, themechanical or obstructive symptoms such as chest pain,aspiration, or shortness of breath are not relieved bymedical therapy. Repair of the giant PEH offers patientsthe opportunity to successfully control both the acid refluxand mechanical symptoms in most cases. Patients appear toexperience early benefit from repair and these benefitsappear to improve further in longer-term follow-up.

It is not surprising that control of symptoms also translatesinto an improvement in QOL. Many studies have focused onQOL as a primary outcome. Previous studies have used ageneric quality of life instrument (SF-36) combined with a

disease specific score such as GERD HQRL1 or a diseasespecific instrument such as QOLRAD2 alone to determinequality of life outcome measures. In our opinion, neither ofthese instruments adequately assesses health in patients withPEH where overall status is impacted by acid reflux,mechanical symptoms, and difficulty with swallowing.Dysphagia is one particular area where QOLRAD andGERD-HRQL are insufficient. Therefore, we have adoptedboth the QOLRAD and GERD-HRQL to assess morecompletely aspects of health and the disease state. We addedthe dysphagia score to directly evaluate swallowing.

The QOLRAD and GERD-HRQL have clearly demon-strated an improvement in quality of life both in the shortand long term compared to their respective preoperativebaseline. However, unlike prior studies, the dysphagia scorehas allowed us to quantify dysphagia and demonstrate thatin short-term follow-up dysphagia is worse, likely due toongoing healing and the reconstruction of the hiatus with itsattendant edema. Dysphagia scores did return to baseline inlong-term follow-up confirming that fear of dysphagiashould not preclude patients from undergoing repair. Thisknowledge allows us to prepare and educate our patientsbefore and early after surgery, counseling them that in mostcases dysphagia if present improves with time.

Observation has been proposed as a reasonable alternativein elderly patients with a minimally symptomatic PEH.12

Using this paradigm, urgent or emergent surgery may berequired if the patient develops rapidly worsening symptomsor acutely incarcerates with or without the presence of agastric volvulus. The nine urgent patients in our series whowere stabilized with nasogastric decompression and earlyendoscopy to rule out strangulation went on to successfullaparoscopic repair with minimal morbidity and restoration ofquality of life. While the ability to help these elderly patientsin an urgent setting is possible with acceptable results, webelieve that decision should be made after surgical consulta-tionwith physicians familiar with treating andmanaging PEH.

The argument against observation is based on otherseries of PEH repairs from respectable centers that havereported an increased mortality and morbidity rate inpatients presenting urgently and undergoing repair.1,3 Whilenot presented in this study, we did exclude from thisanalysis all emergent operations for strangulated PEH,which carries an inherently higher morbidity and mortalityrisk. Our goal should be to avoid observing a patient untilthey present in extremis and require emergent repair. Lastly,the ability to do an appropriate workup for PEH in a stable,elective fashion is far more likely to be successful than afteran urgent admission to the hospital, when tests likemanometry are more difficult to obtain.

The radiographic recurrence rate of 10% compares favor-ably with other series in the literature.6,13 The recurrencesoccurred evenly among our three repair groups suggesting that

J Gastrointest Surg (2011) 15:389396 393

it is likely not the anti-reflux procedure that is central todeveloping a recurrence. It is more likely that adequatemediastinal mobilization of the esophagus and the cruralclosure are central to the outcome of the repair and likelywhether or not a recurrence will develop. Collis gastroplastyhas been proposed as a method to reduce axial tension aftermediastinal mobilization. Even with more liberal use of theCollis gastroplasty, the observed recurrence rates are similar toreports where a Collis was used more selectively. In general,our group advocates using an esophageal lengtheningprocedure selectively. The additional staple lines add in-creased risk for postoperative leak and complications in thisfrail population, but lengthening can be very important if shortesophagus is truly found and there is a need to reduce axialtension on the repair.1,14

Although this study did not focus on the use ofbioabsorbable mesh as an adjunct to the crural repair, weobserved that 50% of our recurrences did not have meshplaced. Since the report by Oelschlager,6 we have changedour practice to reinforce the diaphragm reconstruction withbioabsorbable mesh. There remains some controversy aboutthe utility of mesh1,15 particularly when an esophageallengthening procedure is performed. These two adjuncts toPEH surgery address different physiologic components ofthe pathologic process, namely axial tension (gastroplasty)and radial hiatal tension (mesh). The primary principle inall types of hernia surgery has been to avoid tension.Certainly both adjuncts may be important for an optimaloutcome, as long as the principle of a tension free repair isthe foundation upon which those adjuncts are utilized.

The standard anti-reflux repair associated with PEH repair inNorth America has been the Nissen fundoplication. However, avariety of repairs have been used in reconstruction of thegastroesophageal junction after hiatal closure including partialfundoplication,2,16 Hill repair,17 and the Belsey operation.18

Since both Nissen fundoplication and the Hill repair areperformed at our center, we have observed distinct advantagesand disadvantages to both these operations. To capitalize on theadvantages of both operations (reflux control in the Nissen andaxial maintenance in the Hill), we combined aspects of theseprocedures to see if a hybrid anti-reflux repair would conferdistinct advantages over the traditional repairs (in preparation).8

This study has several strengths and limitations.We believethe use of three different QOL instruments better assesses thequality of life in PEH patients who may have symptoms ofGERD, mechanical symptoms and/or dysphagia. This studydetails consecutive patients 70 years and older undergoingprimarily laparoscopic repair but also includes urgent and opencases. One of the limitations is that dysphagia score used wasnot used in the manner that it was validated. This limits theconclusions we can draw using this score. However, we havefound it to be an important part of our quality of life assessment.Lastly, our median follow-up 1.3 years is short compared to

others even though the range extends out to 5.5 years. We hopeto be able to report on the long-term QOL of this cohort todemonstrate durability of the repair in the future.

Conclusions

These data support repair of symptomatic giant paraesophagealhernias in patients aged 70 years or greater. These hernias canbe repaired in the elderly with minimal surgical mortality andacceptable morbidity in both the elective and urgent setting. Asignificant number of patients undergoing repair can expectresolution of the symptoms they suffered from preoperatively.Similarly, patients should expect improvements in both short-and long-term quality of life measures including patients whopresented urgently or have small recurrent herniation.

References

1. Luketich, J.D., et al., Outcomes after a decade of laparoscopicgiant paraesophageal hernia repair. J Thorac Cardiovasc Surg,2010. 139(2): p. 395404, 404 e1

2. Hazebroek, E.J., et al., Laparoscopic paraesophageal hernia repair:quality of life outcomes in the elderly. Dis Esophagus, 2008. 21(8): p. 73741.

3. Polomsky, M., et al., Should elective repair of intrathoracic stomachbe encouraged? J Gastrointest Surg, 2010. 14(2): p. 20310.

4. Nason, K.S., et al., Laparoscopic repair of giant paraesophagealhernia results in long-term patient satisfaction and a durable repair. JGastrointest Surg, 2008. 12(12): p. 206675; discussion 20757

5. Schieman, C. and S.C. Grondin, Paraesophageal hernia: clinicalpresentation, evaluation, and management controversies. ThoracSurg Clin, 2009. 19(4): p. 47384.

6. Oelschlager, B.K., et al., Biologic prosthesis reduces recurrenceafter laparoscopic paraesophageal hernia repair: a multicenter,prospective, randomized trial. Ann Surg, 2006. 244(4): p. 48190.

7. Aye, R.W., TheHill Procedure for Gastroesophageal Reflux, in CurrentTherapy in Thoracic and Cardiovascular Surgery, S.C. Yang and D.E.Cameron, Editors. 2004, Mosby: Philadelphia, PA. p. 400405.

8. Buduhan, G., et al., The Nissen-Hill "hybrid": Pilot study of a newantireflux repair, in International Society for Diseases of theEsophagus. 2008: Budapest, Hungary.

9. Wiklund, I.K., et al., Quality of Life in Reflux and Dyspepsiapatients. Psychometric documentation of a new disease-specificquestionnaire (QOLRAD). Eur J Surg Suppl, 1998(583): p. 419.

10. Velanovich, V., The development of the GERD-HRQL symptomseverity instrument. Dis Esophagus, 2007. 20(2): p. 1304.

11. Dakkak, M. and J.R. Bennett, A new dysphagia score withobjective validation. J Clin Gastroenterol, 1992. 14(2): p. 99100.

12. Stylopoulos, N., G.S. Gazelle, and D.W. Rattner, Paraesophagealhernias: operation or observation? Ann Surg, 2002. 236(4):p. 492500; discussion 5001.

13. Karmali, S., et al., Primary laparoscopic and open repair ofparaesophageal hernias: a comparison of short-term outcomes.Dis Esophagus, 2008. 21(1): p. 638.

14. Houghton, S.G., et al., Combined transabdominal gastroplasty andfundoplication for shortened esophagus: impact on reflux-relatedand overall quality of life. Ann Thorac Surg, 2008. 85(6): p.194752.

394 J Gastrointest Surg (2011) 15:389396

15. Champion, J.K. and D. Rock, Laparoscopic mesh cruroplasty forlarge paraesophageal hernias. Surg Endosc, 2003. 17(4): p. 5513.

16. Rathore, M.A., et al., Intermediate-term results of laparoscopic repairof giant paraesophageal hernia: lack of follow-up esophagogramleads to detection bias. JSLS, 2007. 11(3): p. 3449.

17. Jobe, B.A., et al., Laparoscopic management of giant type IIIhiatal hernia and short esophagus. Objective follow-up at threeyears. J Gastrointest Surg, 2002. 6(2): p. 1818; discussion 188.

18. Maziak, D.E., T.R. Todd, and F.G. Pearson, Massive hiatus hernia:evaluation and surgical management. J Thorac Cardiovasc Surg,1998. 115(1): p. 5360; discussion 612

Discussant

Dr. Piero Marco Fisichella (Maywood, IL): You show that with anoperation of this high-risk group of patients that can be treated, canachieve good results in terms of quality of life.

However, based on your results, one may think that the operationis still safe. Still, you had a 10% recurrence rate and two perforations.Moreover, the overall complication rate was 24% if you combineminor and major complications. That means that more or less onepatient out of four will have some sort of complication.

I have three questions.First, I am interested in the surgical technique. Based on your

experience, what are the technical elements that can allow you toachieve good results?

You briefly mention in the paper the dissection of the sac, theposterior mediastinal dissection. You also mentioned lengtheningprocedures. Although, you did not use any lengthening procedures, inthe discussion, you say that you used these selectively. In addition,you also said that you used three different techniques.

In summary, could you tell us what is the right approach that youwould use for these patients?

Second. When did recurrence occur? Is there a specific time thatyou saw the recurrence coming? Basically, is there a threshold in thefollow-up beyond which patients may be safe from recurrence?

Third. Do you know if mesh plays a role in the recurrence or not?Last question. You had roughly 25 to 35 patients with short-term

quality of life data, results before and after surgery. And you have68% of patients with long-term results. Do you have any idea what isthe complication rate in these patients?

Closing discussant

Dr. Brian E. Louie: To address your first question around ourtechnique or what we think is important, I think we are like mostlaparoscopic surgeons, we prefer entire sac reduction. We believebringing the sac down is important and detaching it circumferentiallyaround the esophageal hiatus. We spend a considerable amount of timein the operation, probably two thirds of the time mobilizing theintrathoracic esophagus. And our general goal has been to reestablishat least 2 to 3 cm of intra-abdominal esophagus once were satisfiedabout tension.

And if that means taking the dissection up above the inferiorpulmonary veins, that generally means doing so. So we spend aninordinate amount of time doing that. And I think that esophagealmobilization is probably the key to the whole operation. And I think,regardless of which anti-reflux procedure you add on to mobilizationof the esophagus, at least in our series, it doesnt seem to make muchdifference whether we used a Nissen, a Hill, or a hybrid procedure; Ithink mobilization is key.

To answer your third question about the complications and thequality of life and recurrences, the recurrences for us, when we followthese patients, they are generally studied at 6 months and 12 monthswith the barium swallow and/or other tests, so the recurrences generallyoccur between that 6 and 12 month interval. We have seen a couple outlater than that, but I dont have a definite time frame for that.

In terms of quality of life for that group, we didnt pull thatspecifically out for the paper, but the patients that did have theperforations or did get readmission, their general quality of life in thisgroup is generally very good and very similar to the elective group.

And then your other question was recurrence of mesh. So early inthe series, we used no mesh until the report by Dr. Oeschlager andcolleagues saying that mesh reduced the hernia rate, then we began touse mesh much more liberally. Im not sure.

We looked at the data one way and said, you know, we probablyshould be using mesh because of the six recurrences, three didnt havemesh. But the other way to look at it is 60% of our patients didnthave mesh and we still had the same recurrence rates. And I know Dr.Luketichs group said the need for mesh is not as great as everybodythinks it is. I think that is very controversial. For now, I think we aregoing to continue to use mesh.

Discussant

Dr. Nathaniel Soper (Chicago, IL): This is something that we all strugglewith. What do you do with the old patient who has a paraesophagealhernia, because there is a significant morbidity and mortality?

First of all, you state all of these patients had symptoms, so you donot operate on asymptomatic patients who have paraesophagealhernias; is that correct?

Dr. Brian E. Louie: That would not be quite correct because Iwould think we have operated on them. They might not have beenover 70, but in this group they were all symptomatic that werein theconsecutive series, that they all happened to have symptoms.

Discussant

Dr. Nathaniel Soper (Chicago, IL): You said you did not include theemergency operations that were done for strangulation. Just to give usa perspective, in this same period of time, how many of those werethere in your medical center?

Closing Discussant

Dr. Brian E. Louie: In the medical center, we had about a dozen overthe five-year period that the two senior surgeons have counted thatcame in for strangulation and went to the operating room the samenight for endoscopic findings of strangulation, so 12.

Discussant

Dr. Nathaniel Soper (Chicago, IL): And so its so hard to know whatthe denominator is total in any of this series.

Last but not least, you had a 10% recurrence rate, but your meanfollow-up was only about 1.3 years. Do you routinely performanatomical tests to really assess what your true recurrence rate is, orwere these symptomatic patients who happened to get studied?

J Gastrointest Surg (2011) 15:389396 395

Closing discussant

Dr. Brian E. Louie: Our follow-up protocol is generally to get abarium swallow at about a year. And then if the patients are willing,we will undergo full foregut evaluation with endoscopy, pH analysis,

and manometry. We did not include that in this series because we havenot gotten some of the patients out that far yet. But if we follow themlong enough, I think well continue to have objective data onrecurrences down the road.

But it is our protocol generally to get some imaging study, whetherits upper GI esophagogram or an endoscopy.

396 J Gastrointest Surg (2011) 15:389396

2010 SSAT PLENARY PRESENTATION

Preoperative Infliximab is not Associated with an IncreasedRisk of Short-Term Postoperative ComplicationsAfter Restorative Proctocolectomy and IlealPouch-Anal Anastomosis

Melanie L. Gainsbury & Daniel I. Chu & Lauren A. Howard & Jennifer A. Coukos &Francis A. Farraye & Arthur F. Stucchi & James M. Becker

Received: 2 August 2010 /Accepted: 22 October 2010 /Published online: 19 January 2011# 2011 The Society for Surgery of the Alimentary Tract

AbstractIntroduction Considerable controversy exists over whether the preoperative use of infliximab (IFX) for refractory ulcerativecolitis (UC) increases the risk for surgical complications after restorative proctocolectomy and ileal pouch-anal anastomosis(IPAA). The aim of this study was to assess the association between preoperative IFX use and short-term surgicalcomplications in a single-surgeon cohort at a tertiary care academic center.Methods UC patients who underwent IPAA from September 2005 through May 2009 were retrospectively identified.Twenty-nine patients treated with IFX within 12 weeks of surgery and 52 non-IFX control subjects were identified. Short-term postoperative outcomes were compared between groups occurring within 30 days of loop ileostomy closure.Results Patients were similar with respect to demographics, co-morbidities, rate of emergency surgery, hand-sewnanastomosis, and preoperative use of cyclosporine, azathioprine, and high-dose steroids. IFX patients were more likely tohave received a laparoscopic hand-assisted IPAA, low-, medium-, and any-dose steroids, 6-mercaptopurine (6-MP),methotrexate, and to have failed medical therapy. There was no short-term mortality. Overall postoperative and infectiouscomplications were similar between IFX and non-IFX groups. Multivariate regression models revealed no independentpredictors for postoperative complications when including IFX [odds ratio (OR) 0.78, p=0.67], laparoscopic hand-assistedIPAA, 6-MP, methotrexate, steroids, failure of medical therapy, and body mass index.Conclusions Preoperative IFX use was not associated with an increased risk of short-term postoperative complications after IPAA.

Keywords Infliximab . Ulcerative colitis . Ileal pouch-analanastomosis . Short-term complications

Introduction

Ulcerative colitis (UC) is a disease of the colonic mucosacharacterized by recurrent inflammatory episodes. Thetreatment of UC is to a large extent medical, using suchagents as 5-aminosalicylic acid (5-ASA), corticosteroids,and the immunomodulators 6-mercaptopurine (6-MP) andazathioprine. For those patients unresponsive to theaforementioned medications, rescue therapies such ascyclosporine and tumor necrosis factor alpha (TNF-)inhibitors are available.

Approximately one half of chronic UC patients receivingmedical treatment relapse per year. Nearly one fifth of UC

This study was presented, in part, at the 51st Annual Meeting of theSociety for Surgery of the Alimentary Tract in New Orleans, LA onMay 5, 2010 and published in abstract form in Gastroenterology May2010; 138(5):S-867.

Source of financial support The Robert and Dana Smith FamilyFoundation and the Smithwick Endowment Fund, Department ofSurgery, Boston University School of Medicine

M. L. Gainsbury :D. I. Chu : L. A. Howard : J. A. Coukos :A. F. Stucchi : J. M. Becker (*)Department of Surgery, Boston University School of Medicine,88 East Newton Street, C500,Boston, MA 02118, USAe-mail: James.Becker@BMC.org

F. A. FarrayeSection of Gastroenterology,Boston University School of Medicine,Boston, MA, USA

J Gastrointest Surg (2011) 15:397403DOI 10.1007/s11605-010-1385-6

patients experience an acute severe colitis episode requiringhospitalization. Of those, around 60% will respond tointravenous corticosteroids within 72 to 96 h. An additional1520% may improve following rescue medical therapy.Ultimately, however, 30% require surgical management within1 year and 80% will undergo colectomy by 10 years.14

Proctocolectomy with ileal pouch-anal anastomosis (IPAA)remains the surgical procedure of choice for UC patientsrefractory to medical therapy. IPAA offers cure for theintestinal manifestations of the disease while eliminating therisk for colonic malignancy. Recent data from large volumeinstitutions suggest improved health-related quality of lifefollowing IPAA with reliable functional outcomes.4

Infliximab (IFX) is a chimeric IgG1 monoclonal anti-body that targets TNF-, an important regulator of manychronic inflammatory diseases such as UC.5 IFX receivedFDA approval in September 2005 for use in induction andmaintenance therapy for moderate to severe UC. AlthoughIFX holds several boxed warnings, including the increasedrisk of malignancy and opportunistic infections such asdisseminated fungal infections and tuberculosis, its use isconsidered safe and effective. Recent studies, however,have shown that between 30% and 50% of patients treatedwith IFX still fail rescue therapy and proceed to colec-tomy.4 Considerable controversy exists in the literature onwhether such preoperative IFX use increases short-termpostoperative complications for these patients after procto-colectomy with IPAA.

As summarized in Table 1, a study by Selvasekar et al. atthe Mayo Clinic found that IFX use in UC patients within2 months of IPAA significantly increased the risk foranastomotic leak, pouch-specific, and infectious complica-tions.6 Similarly, Mor et al. from the Cleveland Clinicreported significantly increased rates of anastomotic leak,pouchitis, abscess, and overall complications in UC andindeterminate colitis patients with preoperative IFX expo-sure.7 Conversely, Kunitake et al. at Massachusetts GeneralHospital found similar rates of pouch-specific, surgery-related, and infectious complications between preoperativeIFX and non-IFX groups.8 These results were supported byFerrante et al. who found no differences in anastomotic leak,

pelvic abscess, pouch-related or infectious complications intheir study population.9 A recent meta-analysis by Yang et al.further confounds the literature by reporting an associationbetween IFX exposure and overall postoperative complica-tions but no association individually between preoperativeIFX use and infectious or non-infectious complications forUC patients.10 These studies demonstrate that the surgicalcommunity is still unclear as to whether patients whoundergo medical rescue therapy with IFX prior to an IPAAcan expect a safe and functional outcome.

Therefore, the aim of our study was to assess theassociation between preoperative IFX use and short-termsurgical complications in a single-surgeon cohort at ourtertiary care academic referral center.

Methods

Data were collected from an IRB-approved IPAA Registryat Boston University Medical Center. We identified 81consecutive UC patients who underwent IPAA betweenSeptember 2005 and May 2009 by a single surgeon (J.M.B.).Of the 81 subjects, 29 had received IFX treatment within12 weeks of the first stage of their IPAA surgery. Fifty-twocontrol subjects remained as the non-IFX group. Short-termpostoperative outcomes were compared between the twogroups as described below.

Inclusion and Exclusion Criteria

Patients with a diagnosis of UC registered in the IPAAdatabase who underwent IPAA at Boston University MedicalCenter between September 2005 and May 2009 wereincluded in this study. Patients with other pre- or postoper-ative diagnoses such as Crohns disease, familial adenoma-tous polyposis, or indeterminate colitis were excluded.

Clinical Variables

Medical records of all included subjects were retrospec-tively reviewed. Data abstracted from the medical recordsincluded the following patient demographics: age, gender,body mass index (BMI), smoking status, American Societyof Anesthesiologists (ASA) class, and co-morbidities.Information regarding medications used 12 weeks prior tosurgery included IFX, cyclosporine, methotrexate, 6-MP,azathioprine, oral low-dose steroids (40 mg/day). Surgical factors evaluated includedindication for surgery, type of procedure (two- versus three-stage), modality (open versus laparoscopic hand-assisted),and ileal pouch-anal anastomosis technique (stapled versushand-sewn).

Table 1 Literature-based comparison of postoperative complicationrisk associated with preoperative use of infliximab

Authors Overall complicationsOR (95% CI)

Infectious complicationsOR (95% CI)

Selvasekar et al.6 1.7 (0.93.2) 2.7 (1.16.7)

Mor et al.7 3.5 (1.58.3) 13.8 (1.8105)

Kunitake et al.8 1.1 (0.62.0) 0.5 (0.21.4)

Ferrante et al.9 Not available 0.3 (0.071.4)

Yang et al.10 1.8 (1.12.9) 2.2 (0.68.0)

398 J Gastrointest Surg (2011) 15:397403

Outcome Measures

Our primary outcome was the rate of overall short-termpostoperative complications between the IFX and non-IFXgroups. Secondary outcomes included the rate of short-terminfectious and non-infectious complications. Short-term post-operative complications were defined as having occurredbetween the first-stage IPAA surgery until up to within 30 daysafter the last-stage IPAA surgery, the closure of the divertingloop ileostomy. Complications were defined as pouch/anasto-motic leak, pelvic/intraabdominal abscess, pouch-related com-plications, wound infection, and other. Other complicationsincluded thrombosis (pulmonary embolus, portal vein throm-bosis, and deep venous thrombosis), small bowel obstruction,ileus, and one episode of wound dehiscence. Pouch-relatedcomplications included one episode of pouch dehiscence andone episode of a fistula originating from the pouch. Pouch oranastomotic leaks were defined as contrast extravasations seenon computed tomography or loop-o-gram studies. Woundinfections occurring in the midline, port-site, or ostomy-sitelocations were included in this study.

Statistical Analysis

Categorical variables were reported as frequencies andpercentages. Continuous variables were reported as meanand standard deviation (SD). Students t test for continuousvariables and chi-square or Fishers exact tests for categor-ical variables were used as appropriate in evaluating theassociations between IFX use and patient factors such asdemographics, medications, and IPAA data. Logisticregression analysis was used to assess multivariableassociations between potential risk factors and the follow-ing outcomes: overall postoperative complications, infec-tious and non-infectious complications, and woundinfection. Results are presented as odds ratios (OR) with95% confidence intervals (CI).

A p value of 40 mg/day) use wereobserved (Table 2). Infliximab patients, however, were morelikely to have received 6-MP, methotrexate, low-dose steroids(

assisted IPAA, BMI, and use of any-dose steroid, 6-MP, ormethotrexate (Table 4). Logistic regression models alsorevealed no independent predictors of infectious or non-infectious complications when including these same factors(Table 4). On multivariate logistic regression, patients weremore likely to develop wound infections with higher BMIs(OR 0.88, CI 0.780.99, p=0.049) (Table 5). IFX, any-dosesteroids, 6-MP, failure of medical therapy, and laparoscopichand-assisted procedures were not found to be predictors ofwound infection (Table 5).

Subgroup Analysis

In a subgroup analysis in which all urgent/emergency andthree-stage IPAA surgery patients were excluded, theresults remained very similar. Logistic regression modelscontinued to reveal no independent predictors of overall,infectious, or non-infectious complications when includingIFX. On multivariate logistic regression, however, BMI nolonger predicted the development of wound infections (OR0.89, CI 0.781.01, p=0.06).

Complication IFX (n=29) Non-IFX (n=52) p value

Overall 13 (44.8%) 23 (44.2%) 0.96

Infectious 5 (17.2%) 14 (26.9%) 0.32

Pelvic/intraabdominal abscess 4 (13.8%) 7 (13.5%) 1.00

Wound infection 1 (3.5%) 10 (19.2%) 0.09

Non-infectious 12 (41.4%) 16 (30.8%) 0.34

Pouch/anastomotic leak 1 (3.5%) 5 (9.6%) 0.41

Pouch-related 0 (0.0%) 2 (3.9%) 0.53

Other 12 (41.4%) 13 (25.0%) 0.13

Table 3 Short-term complicationrates compared between inflixi-mab and non-infliximab groups

Demographics IFX (n=29) Non-IFX (n=52) p value

Patient factors

Age, yearsa 36.212.6 42.012.7 0.06

BMIa 27.07.0 27.65.9 0.68

Gender, male 11 (37.9%) 22 (42.3%) 0.70

ASA score 2 21 (72.4%) 44 (84.6%) 0.68Smoker 5 (17.2%) 18 (34.6%) 0.10

Co-morbidities

Diabetes mellitus 2 (6.9%) 1 (1.9%) 0.29

Hypertension 6 (20.7%) 4 (7.7%) 0.16

Cardiac 4 (13.8%) 1 (1.9%) 0.06

Pulmonary 2 (6.9%) 4 (7.7%) 1.00

Renal 0 (0%) 3 (5.8%) 0.55

Surgical factors

Failed medical therapy 26 (89.7%) 36 (69.2%) 0.04

Emergent/urgent first stage 3 (10.3%) 5 (9.6%) 1.00

2-stage IPAA 28 (96.6%) 47 (90.4%) 0.41

Laparoscopic colectomy 13 (44.8%) 4 (7.7%)

Discussion

Our study indicates that preoperative IFX use 12 weeks prior toundergoing IPAA for UC is not associated with an increasedrisk of overall short-term postoperative complications. More-over, no differences were observed in infectious or non-infectious complications between IFX- and non-IFX-treatedpatients. These findings suggest that for UC patients refractoryto medical therapy, a rescue trial of IFX will not affect theshort-term postoperative outcomes for those who subsequentlyrequire restorative proctocolectomy and IPAA. To our knowl-edge, we are the first study to examine only patients who hadreceived IFX after its FDA approval in September 2005 for usein moderate to severe UC. Prior studies have included patientswho received IFX during off-label usage, which can makeresults difficult to interpret as these patients may have been inpoorer condition prior to surgery.

Infliximab patients were more likely to have failed medicaltherapy and to have received methotrexate, 6-MP, and low-,medium-, and any-dose steroids. These observations were notsurprising since IFX use is generally reserved for those patientsfailing other medical therapies. Ultimately our data show noincreased risk among IFX exposed patients for overall,infectious, or non-infectious complications. Of note, the

proportion of ASA scores 2 at first-stage IPAA were similaramong patients with an overall complication and those withoutany complication (p=1.00, data not shown). We therefore donot believe that overall health status is confounding thelikelihood of developing a complication after surgery. In ourstudy population, IFX patients were more likely to haveundergone laparoscopic hand-assisted IPAA. This is anunusual association not previously reported in the literature.Our institution does not have any preset selection criteria forlaparoscopic hand-assisted procedures, which makes thisfinding difficult to reconcile. Incidentally, a recent studysuggests preoperative IFX treatment does not affect outcomesafter laparoscopic restorative proctocolectomy with IPAA.11

Interestingly, our data demonstrated a trend toward fewerwound infections in the IFX treated group, however, this wasnot statistically significant. Regression analysis revealed higherBMIs to be predictive of developingwound infections, which iswidely supported by the literature.1215 Logistic regression didnot show laparoscopic hand-assisted proctocolectomy to beprotective against wound infection in our study populationdespite reports in the literature suggesting the contrary.16,17

Our findings are incongruent from those of Mor et al.7

and Selvasekar et al.,6 who reported increased risk ofpostoperative complications after IFX use. In the Mor et

Table 4 Multivariate logistic regression analysis of factors associated with postoperative complications after IPAA

Covariate Overall complication Infectious complication Non-infectious complication

IFXa 0.78 (0.262.38) p=0.67 1.87 (0.467.57) p=0.38 0.59 (0.191.87) p=0.37

Steroid, any 1.29 (0.325.29) 2.41 (0.4612.7) 1.02 (0.234.46)

6-MP 1.05 (0.382.89) 1.02 (0.303.54) 0.81 (0.282.33)

Methotrexate 2.43 (0.2030.1) NAb 1.79 (0.1423.0)

Failed medical therapy 0.94 (0.243.61) 0.57 (0.113.03) 1.41 (0.355.65)

Laparoscopic colectomy 1.25 (0.305.10) 0.31 (0.061.72) 1.13 (0.274.82)

BMI 1.02 (0.941.10) 0.93 (0.851.03) 1.04 (0.951.14)

Results are expressed such that OR 1.0 predicts the absence of the outcomea Results expressed as odds ratio, confidence interval, and p value. All other results expressed as odds ratio and confidence intervalb Due to the presence of zero cells, logistic regression for methotrexate is not valid

Table 5 Multivariate logistic regression analysis of factors associated with wound infection after IPAA

Covariate Odds ratio 95% Confidence interval p value

IFX 9.49 0.71126.6 0.09

Steroid, any 9.47 0.9396.2 0.06

6-MP 0.36 0.061.98 0.24

Methotrexatea NA NA NA

Failed medical therapy 0.21 0.022.42 0.21

Laparoscopic colectomy 0.31 0.024.73 0.40

BMI 0.88 0.780.99 0.049

Results are expressed such that OR 1.0 predicts the absence of the outcomea Due to the presence of zero cells, logistic regression for methotrexate is not valid

J Gastrointest Surg (2011) 15:397403 401

al.7 study, immunomodulators were more frequently usedamong the IFX-treated group. Immunomodulator use wasone of the factors adjusted for on multivariate analysis,which greatly minimizes but can never completely elimi-nate its influence on study results. They also looked atpatients with any preoperative exposure to IFX, with a 37-week upper interquartile range. Upon subset analysis, theauthors reported that whether patients received IFX within16 weeks of their surgery or after did not change the factthat sepsis was significantly greater in the IFX group. Theduration of infliximabs biological activity is not known,but with a half-life of 7 to 12 days and onset of action ofapproximately 2 weeks, it would seem unlikely that IFXalone could be responsible for this increased risk after16 weeks. In the Selvasekar study,6 IFX patients were morelikely to be on high-dose steroids, 5-ASA, and azathioprine,which were also adjusted for on multivariate analysis.

In the studies by Kunitake et al.8 and Ferrante et al.,9 IFXwas not found to increase the risk of postoperative complica-tions. Kunitake et al.8 investigated UC, Crohns disease, andindeterminate colitis patients undergoing any abdominalsurgery. The results of a mixed IBD cohort may be difficultto interpret since Crohns patients do not seem to be atincreased risk for postoperative complications from IFX.1820

In the study by Ferrante et al.,9 IFX-exposed patients wereyounger, had shorter disease duration prior to surgery, andlower C-reactive protein levels. Although these factors wereexamined on univariate analysis, multivariate analysis was notperformed. One could argue that these patients were healthierthan their control counterparts, and perhaps less likely todevelop complications. The Ferrante et al.9 study populationincluded patients who underwent a single-stage IPAAwithoutan ileostomy. It is therefore difficult to reconcile results withthose at our institution, where two- or three-stage procedureswith diverting loop ileostomies are consistently performed.Furthermore, the authors reported IFX-exposed patients weremore likely to receive an IPAA with ileostomy than controls.And the patient cohort without ileostomies was found atincreased risk for complications, further confounding the data.

Beyond UC, infliximab has been used in several otherpreoperative clinical settings. There seems to be consensus inthe literature regarding Crohns disease, as several studies haveshown no increased postoperative risk associated with IFXuse.1820 Controversy is, however, apparent in orthopedicliterature. Giles et al. reported increased risk of infectiouscomplications following orthopedic procedures in rheumatoidarthritis patients on IFX.21 Conversely, others have found nosuch increased risk among IFX-exposed rheumatoid arthritispatients after orthopedic surgery.22,23

Our study is not without its limitations. It is retrospectivelydesigned, examines a single center study population operatedon by a single surgeon, and has a small sample size. It ispossible that our study is not powered enough to detect small

differences between the IFX and non-IFX groups. Laparo-scopic hand-assisted IPAA, failure of medical therapy,low-, medium-, and any-dose steroid use, methotrexate use,and 6-MP use were unequally distributed among the groups.The effects of these differences were minimized by inclusionin multivariate analysis models but never eliminated. Further-more, there may be other potential confounders we wereunable to assess such as UC disease severity, malnutrition,duration of colitis prior to surgery, and total number of IFXinfusions received. Other factors associated with septiccomplications following IPAA have recently been reportedby the Cleveland Clinic including BMI, blood transfusion,and individual surgeon. This study was unable to find anassociation between IFX use and septic outcomes.24

No study should be taken in isolation. The need for amulti-centered prospective study or a collaborative retro-spective study from multiple registries with well-definedvariables echoed in the literature deserves to be re-mentioned. An end to the controversy regarding IFX usein UC and postoperative complications is unlikely to befound without such an undertaking.

Financial disclosures Dr. Farraye sits on the Advisory Board forCentocor and recently resigned from Centocor's Speakers Bureau 8/2010.

Acknowledgement The authors would like to acknowledge JeremyHetzel for his statistical advice.

References

1. Langholz E, Munkholm P, Davidsen M, Binder V. Course ofulcerative colitis: analysis of changes in disease activity overyears. Gastroenterology 1992; 103:14441451

2. Baudet A, Rahmi G, Bretagne AL, et al. Severe ulcerative colitis:present medical treatment strategies. Expert Opin Pharmacother.2008; 9:447457.

3. Becker JM, Stucchi AF. Treatment of choice for acute severesteroid-refractory ulcerative colitis is colectomy. Inflamm BowelDis 2009; 15(1):146149

4. Becker JM, Stucchi AF. Is surgery the best second-line therapy inacute ulcerative colitis? A pro-surgery viewpont. Inflamm BowelDis Monitor 2007; 8(2):4959

5. Tracey KJ, Cerami A. Tumour necrosis factor: a pleiotropiccytokine and therapeutic target. Annu Rev Med 1994; 45:491503

6. Selvasekar CR, Cima RR, Larson DW, Dozois EJ, et al. Effect ofinfliximab on short-term complications in patients undergoingoperation for chronic ulcerative colitis. J Am Coll Surg 2007;204:956963

7. Mor IJ, Vogel JD, Moreira AL, Shen B, et al. Infliximab inulcerative colitis is associated with an increased risk of postop-erative complications after restorative proctocolectomy. Dis ColonRectum 2008; 51:12021210

8. Kunitake H, Hodin R, Shellito PC, Sands BE, et al. Perioperativetreatment with infliximab in patients with Crohns disease andulcerative colitis is not associated with an increased rate ofpostoperative complications. J Gastrointest Surg 2008; 12:17301737

402 J Gastrointest Surg (2011) 15:397403

9. Ferrante M, DHoore A, Vermeire S, Declerck S, et al. Cortico-steroids but not infliximab increases short-term postoperativeinfectious complications in patients with ulcerative colitis. InflammBowel Dis 2009; 15:10621070

10. Yang Z,Wu Q,WuK, Fan D.Meta-analysis: pre-operative infliximabtreatment and short-term post-operative complications in patients withulcerative colitis. Aliment Pharmacol Ther 2010; 31:486492

11. Reinier BC, Berdah SV, Grimaud JC, et al. Preoperativeinfliximab treatment and postoperative complications after lapa-roscopic restorative proctocolectomy with ileal pouch-anal anas-tomosis: a case-matched study. Surg Endosc 2010; 24:18661871

12. Nagachinta T, Stephens M, Reitz B, Polk F. Risk factors forsurgical-wound infection following cardiac surgery. J Infect Dis1987; 156(6):967973

13. Tran TS, Jamulitrat S, Chongsuvivatwong V, Geater A. Riskfactors for postcesarean surgical site infection. Obstet Gynecol2000; 95(3):367371

14. Dindo D, Muller MK, Weber M, Clavien PA. Obesity in generalelective surgery. Lancet 2003; 361:20322035

15. Pikarsky AJ, Saida Y, Yamaguchi T, Martinez S, et al. Is obesity ahigh-risk factor for laparoscopic colorectal surgery? Surg Endos-copy 2002; 16(5):855858

16. Varela JE, Wilson SE, Nguyen NT. Laparoscopic surgerysignificantly reduces surgical-site infections compared with opensurgery. Surg Endosc 2010; 24(2):270276

17. Romy S, Eisenring MC, Bettschart V, et al. Laparoscopic use andsurgical site infections in digestive surgery. Ann Surg 2008; 247(4):627632

18. Colombel JF, Loftus EV Jr, Tremaine WJ et al. Early postoper-ative complications are not increased in patients with Crohnsdisease treated perioperatively with infliximab or immunosup-pressive therapy. Am J Gastroenterol 2004; 99:878883

19. Marchal L, DHaens G, Van Assche G, et al. The risk of post-operative complications associated with infliximab therapy forCohns disease: a controlled cohort study. Aliment PharmacolTher 2004; 19:749754

20. Lichtenstein GR, Feagan BG, Cohen RD, et al. Serious infectionsand mortality in association with therapies for Crohns disease:TREAT registry. Clin Gastroenterol Hepatol 2006; 4(5):621630

21. Giles JT, Bartlett SJ, Gelber AC, et al. Tumour necrosis factor inhibitortherapy and risk of serious postoperative orthopedic infection inrheumatoid arthritis. Arthritis Rheum 2006; 55:333337

22. Bibbo C, Goldberg JW. Infectious and healing complications afterelective orthopaedic foot and ankle surgery during tumour necrosisfactor-alpha inhibition therapy. Foot Ankle Int 2004; 25(5):331335

23. den Broeder AA, Creemers MC, Fransen J, et al. Risk factors forsurgical site infections and other complications in elective surgeryin patients with rheumatoid arthritis with special attention for anti-tumor necrosis factor: a large retrospective study. J Rheumatol2007; 34(4):689695

24. Kiran RP, Moreira AL, Remzi FH, Church JM, et al. Factorsassociated with septic complications after restorative proctocolec-tomy. Annals Surgery 2010; 251(3):436440

Discussant

Dr. Amy L. Halverson (Chicago, IL): I congratulate youon taking on this controversial topic. And really one of themost important issues is what are the baseline patientcharacteristics in terms of impacting outcome versus whatis the impact of the infliximab. That is, and is the use ofinfliximab just a marker for more severe disease?

I want to focus on two questions.The first question involves looking at the patients that

underwent the initial ileal pouch operation versus thosepatients that were so sick they were deemed to undergo justthe colectomy and then have an interval ileal pouchoperation. Now, I noticed that in your control group, therewere some patients that underwent just a colectomy and asubsequent ileal pouch. In contrast, with the infliximabgroup, they all underwent the ileal pouch initially.

So can you talk a little bit about how you decide whogets the pouch and who gets a colectomy and then a pouch,and then how you think that those patients that underwentthe colectomy and then the subsequent pouch affect theoutcomes? Do you think that maybe they are at increasedrisk for complications because they were sicker, or do youthink that they are at decreased risk because they had theirpouch surgery long after the other morbidity related to theulcerative colitis was sort of out of the picture and they hada more elective pouch operation?

My second question relates to this laparoscopic surgery.Can you give a little insight into how you think that there issuch a difference in the infliximab versus the non-infliximabgroup and the role that laparoscopic surgery plays with that?

Closing Discussant

Dr. Melanie L. Gainsbury: To address the first question,certainly patients who underwent an emergency first-stageprocedure, causing them to have a three-staged IPAA,were verydifferent from their two-staged counterparts. Those requiringemergency surgery are certainly much sicker and unable totolerate the pouch creation at the time of their first surgery.Whether inclusion of these patients would impact the databecause they are at increased risk for complications or perhapsat less risk because they were staged was difficult to tell.

But wewere able to actually run a subset analysis where weeliminated those patients who were emergency surgeries, andwe did not find significant changes in the data. Essentially, therate of overall complications, infectious complications, andnon-infectious complications between the infliximab and non-infliximab groups continued to be insignificantly different.

In terms of the second question, regarding the laparoscopiccolectomies, it was rather surprising to us at first when wediscovered that the rate of laparoscopic colectomies wassignificantly different between the infliximab and non-infliximab groups. We did not have any preset criteria at ourinstitution for selecting patients for laparoscopic colectomy,so it was rather difficult for us to reconcile this difference. Weincluded it as a factor in all of our multivariate analyses to tryto help offset some of that influence on the data.

J Gastrointest Surg (2011) 15:397403 403

SSAT/AHPBA JOINT SYMPOSIUM 2010

SSAT/AHPBA Joint Symposium: Today's Approachesto Colorectal Cancer (CRC) Liver Metastases

W. Scott Helton

Received: 2 January 2011 /Accepted: 11 January 2011 /Published online: 1 February 2011# 2011 The Society for Surgery of the Alimentary Tract

Each year in the USA, approximately 150,000 patients arediagnosed with colorectal cancer with an associated 55,000attributable deaths.1 Colorectal cancer is the second mostcommon cause of cancer-related death in the USA, withmost patients dying of metastatic disease. Up to 4050% ofpatients with colorectal cancer will develop metastasis,2

with about 1020% presenting with liver metastasis(CRLM) at the time of diagnosis3,4 and another 2025%developing metachronous liver metastasis some timelater.5,6 Despite this high incidence of developing meta-static disease, the median survival of patients with CRLMhas increased substantially over the past 10 years; patientsare living with persistent or recurrent metastatic diseaselonger and with better quality of life than those treated inthe 1980s and 1990s.7,8 This improvement in outcome isrelated to improved patient selection, newer and moreaggressive surgical techniques,9,10 and more effectivechemotherapy agents and regimens.11 Five-year survivalfollowing curative intent surgery of CRLM now approaches4560%1215 and is as high as 20% at 10 years in selectindividuals.16

Patients with extensive metastatic liver disease previous-ly thought to be unresectable can now be rendered free ofdisease after receiving multimodality therapy that includesboth systemic and liver-directed therapies. Modern liver-directed therapy may include simple or radical liver

resection, and second- and third-stage hepatectomies aswell as a variety of nonresection therapies includingablation (radiofrequency, microwave, cryoablation, electro-poration), radiotherapy (external beam or transvascular),and hepatic artery infusion chemotherapy.17 The laterapproaches can be employed singularly or in combinationwith resection and systemic therapy to reduce and/orcontrol the magnitude of metastatic disease and rendermany patients completely free of disease.

Despite these advances, many patients with CRLM whocould potentially benefit from multidisciplinary liver-directed therapy alone or in combination with neoadjuvantand adjuvant therapy are not seen by clinicians who areknowledgeable and experienced with CRLM and hence arenot offered this opportunity for improved survival. The lackof understanding by medical oncologists, gastroenterolo-gists, and many surgeons on how liver-directed therapies fitin with modern chemo and biological regimens promptedthe American Hepatopancreatobiliary Association(AHPBA), the Society of Surgical Oncology (SSO), andthe Society for Surgery of the Alimentary Tract (SSAT) tohost a consensus conference on this topic in 2006 duringthe Annual Meeting of the American Society of ClinicalOncology (ASCO) meeting.18,19 Since 2006, additionalexperience has been accrued with more widespread aggres-sive multimodality therapy to convert patients fromunresectable to resectable status. Kopetz and colleaguespublished a review in The Lancet that discussed this newparadigm for patients with both resectable and initiallyunresectable disease (see Fig. 1).20 Despite these efforts ofknowledge dissemination, the rapid evolution of so manynew approaches and therapies has confused many practi-tioners and patients. The rapid evolution of therapy andpaucity of adequately powered randomized clinical trials todefine best therapy have led patients and physicians to ask a

Presented at SSAT/AHPBA Joint Symposium: Today's Approaches toColorectal Cancer Liver Metastases, Digestive Disease Week, NewOrleans, LA 2010

W. S. Helton (*)Department of General, Vascular & Thoracic Surgery,Virginia Mason Medical Center,1100 9th Ave,Seattle, WA 98101, USAe-mail: scott.Helton@vmmc.org

J Gastrointest Surg (2011) 15:404405DOI 10.1007/s11605-011-1425-x

number of questions surrounding therapeutic choices.These include but are not limited to:

& Can we currently identify individuals who will benefitfrom specific therapies and what factors, if any, arereliable predictors of survival?

& With so many therapies, which one is best for whichpatient?

& If chemotherapy is so much better today for CRLM,should everyone receive it, regardless of havingresectable liver tumor(s) upon presentation? If so,should they receive chemotherapy before or after liverresection? For those who receive chemotherapy beforeliver resection, does this increase subsequent surgicalmorbidity or even preclude surgery?

& For patients who refuse or are ineligible for liver resection,which other therapies are most effective or reasonable?

& What are the limits of liver resection today and howmuch residual liver reserve is necessary for survival?

& How do we balance the benefits and risks of resectionagainst other less invasive therapies for a given patient?

In an effort to address these questions, the scientific andprogram committees of the AHPBA and SSAT agreed thatit would be worthwhile to host a symposium on themanagement of CRLM during Digestive Disease Week inthe Spring of 2010. The goals of this symposium were toincrease knowledge of surgeons and gastroenterologistsabout recommended best practice based upon currentevidence for treating patients with CRLM and to bringclarity to many of the questions above. Four experiencedsurgical oncologist from four prominent cancer centersdelivered outstanding talks on this subject. The AHPBAand SSAT are extremely pleased and appreciative of the factthat they agreed to publish their comments in the Journal of

Gastrointestinal Surgery. With this effort, our two associ-ations hope to advance the care and improve the outcomesof patients with CRLM.

References

1. Jemal A, Murray T, Ward E, et al: Cancer statistics, 2005. CACancer J Clin 55:1030, 2005

2. Steele G, Jr., Ravikumar TS: Resection of hepatic metastases fromcolorectal cancer. Biologic perspective. Ann Surg 210:12738, 1989

3. Cady B, Monson DO, Swinton NW: Survival of patients aftercolonic resection for carcinoma with simultaneous liver metasta-ses. Surg Gynecol Obstet 131:697700, 1970

4. Jatzko G, Wette V, Muller M, et al: Simultaneous resection ofcolorectal carcinoma and synchronous liver metastases in a districthospital. Int J Colorectal Dis 6:1114, 1991

5. Scheele J, Stang R, Altendorf-Hofmann A, et al: Resection ofcolorectal liver metastases. World J Surg 19:5971, 1995

6. Altendorf-Hofmann A, Scheele J: A critical review of the majorindicators of prognosis after resection of hepatic metastases fromcolorectal carcinoma. Surg Oncol Clin N Am 12:16592, xi, 200

7. Adson MA, van Heerden JA, Adson MH, et al: Resection of hepaticmetastases from colorectal cancer. Arch Surg 119:64751, 1984

8. Hughes KS, Rosenstein RB, Songhorabodi S, et al: Resection ofthe liver for colorectal carcinoma metastases. A multi-institutionalstudy of long-term survivors. Dis Colon Rectum 31:14, 1988

9. Abdalla EK, Resection of colorectal liver metastases. Currentarticle in J Gastrointestinal Surgery, 2010.

10. Chun YS, Vauthey JN, Ribero D, et al. Systemic chemotherapyand two-stage hepatectomy for extensive bilateral colorectal livermetastases: perioperative safety and survival. J Gastrointest Surg.2007 Nov;11(11):1498504.

11. Pawlik TM, Cosgrove D. The role of peri-operative chemotherapyfor resectable colorectal liver mestastsis: what does the evidencesupport? Current article in J Gastrointestinal Surgery, 2010.

12. Abdalla EK, Vauthey JN, Ellis LM, et al: Recurrence andoutcomes following hepatic resection, radiofrequency ablation,and combined resection/ablation for colorectal liver metastases.Ann Surg 239:81825; discussion 8257, 2004

13. Choti MA, Sitzmann JV, Tiburi MF, et al: Trends in long-termsurvival following liver resection for hepatic colorectal metasta-ses. Ann Surg 235:75966, 2002

14. Pawlik TM, Scoggins CR, Zorzi D, et al: Effect of surgical marginstatus on survival and site of recurrence after hepatic resection forcolorectal metastases. Ann Surg 241:71522, discussion 7224, 2005

15. de Jong MC, Mayo SC, Pulitano C, et al: Repeat curative intentliver surgery is safe and effective for recurrent colorectal livermetastasis: results from an international multi-institutional analy-sis. J Gastrointest Surg 13:214151, 2009

16. Tomlinson JS, Jarnagin WR, DeMatteo RP, et al: Actual 10-yearsurvival after resection of colorectal liver metastases defines cure.J Clin Oncol 25:457580, 2007

17. Boutros C, Espat NJ, What how and when to offer non-resectionaltherapy for colorectal cancer liver metastases. Current article in JGastrointestinal Surgery, 2010.

18. Bartlett DL, Berlin J, Lauwers GY, Messersmith WA, Petrelli NJ,Venook AP. Ann Surg Oncol. 2006 Chemotherapy and regionaltherapy of hepatic colorectal metastases: expert consensusstatement. Oct;13(10):128492. Epub 2006 Sep 6

19. JN Vauthey, M Choti, WS Helton. AHPBA/SSO/SSAT ConsensusConference on Hepatic Colorectal Metastases: Rationale and Over-view of the conference.. Ann Surg Oncol, 2006. Oct;13(10):125960

20. Kopetz S, Vauthey JN. Perioperative chemotherapy for resectablehepatic metastases. Lancet, 2008 Mar 22:371(9617):100716.

A New Paradigm for Colorectal Liver Metastases

Resectable Unresectable

Hepatectomy (One-stage or

Two-stage) PVE*

Preoperative Therapy

2-3 months First-Line

Chemotherapy Re-evaluate 2-3 months

Second-Line Chemotherapy

Third-Line Chemotherapy

Resectable

Postoperative Therapy

3-4 months

+/- liver-directed therapy

+/- liver-directed therapy

Fig. 1 A new paradigm for colorectal liver metastases. *Portal VeinEmbolization

J Gastrointest Surg (2011) 15:404405 405

SSAT/AHPBA JOINT SYMPOSIUM 2010

Prognostic Markers and Staging Systems for Patientswith Colorectal Liver Metastases

James J. Mezhir & Michael I. DAngelica

Received: 2 January 2011 /Accepted: 11 January 2011 /Published online: 29 January 2011# 2011 The Society for Surgery of the Alimentary Tract

Keywords Colorectal cancer . Hepatic metastasis .

Chemotherapy . Clinical risk score . Biomarkers

AbbreviationsCRLM Colorectal liver metastasesCEA Carcinoembryonic antigenCRS Clinical risk score

Introduction

The treatment of colorectal liver metastasis has evolvedover the last 20 years. What was previously thought to be acontraindication to surgery, metastatic disease in the liverhas been demonstrated to be amenable to locoregionaltherapy. Up to 70% of recurrences after resection ofprimary colorectal cancer occur in the liver and up to40% of these patients present with liver-only disease. Thesepatients are amenable to potential cure following metasta-sectomy as demonstrated by 10 year actual survival ofnearly 20% in selected patients.1

The treatment of colorectal liver metastasis (CRLM) ismultidisciplinary and often includes treatment withregional or systemic chemotherapy. The use of chemo-therapy as an adjunct to surgical resection, however, is

debatable and the therapeutic impact is relatively small.The optimal timing (before or after surgery) and durationof treatment is unknown. To help guide the clinicianwhen evaluating patients with CRLM, numerous groupshave developed prognostic scoring systems based onretrospective analyses.25 These scoring systems includeclinicopathologic factors that impact outcome such asnodal disease in the primary tumor, timing of thedevelopment of metastasis (synchronous vs. metachro-nous), size and number of metastasis in the liver,carcinoembryonic antigen (CEA) level, and the presenceof extrahepatic disease.

Despite these efforts, currently there is no idealpredictor of outcome for patients with resectable CRLM.The ideal predictor of outcome would include thefollowing characteristics: low cost and easy to measure,reproducible across institutions, and measurable bothbefore and after treatment. Most importantly, this factorwould predict major differences in outcome that signif-icantly impact treatment (Fig. 1a). A clinical example ofthis paradigm is K-ras status as a predictor of response totherapy with cetuximab, a monoclonal antibody againstthe epidermal growth factor receptor.6 In a prospectiverandomized controlled trial, patients with advanced colo-rectal cancer were randomized to treatment with orwithout cetuximab. When stratified for K-ras status,patients with wild type K-ras tumors demonstrated asignificant survival advantage compared to those withmutated K-ras tumors, who derived no benefit from thechemotherapeutic agent (Fig. 1b). Therefore, patients withmutated K-ras do not receive cetuximab therapy and arespared the toxicity associated with a treatment with noproven benefit. To date, there is no specific clinical risk

J. J. Mezhir :M. I. DAngelica (*)Department of Surgery, Section of HepatopancreaticobiliarySurgery, Memorial Sloan-Kettering Cancer Center,1275 York Avenue,New York, NY 10065, USAe-mail: dangelim@mskcc.org

J Gastrointest Surg (2011) 15:406409DOI 10.1007/s11605-011-1424-y

score or biomarker that specifically prognosticates orguides therapy for patients with resectable CRLM to thisdegree.

Clinical risk scoring systems are based on multivariateanalyses of large clinical databases of patients selected forsurgical resection. Multiple independent predictive factorsare combined into a score that correlates with outcome.The Memorial Sloan Kettering Cancer Center clinical riskscore (CRS) was established after review of 1,001consecutive hepatic resections for CRLM.3 The followingpreoperative factors were significant predictors of disease-free survival on multivariate analysis (1 point is earned foreach factor): node-positive primary tumor, disease freeinterval 5 cm, CEA >200 ng/mL. A score of 0 wasassociated with a 5-year recurrence-free survival (RFS) of60% vs. a score of 5 which was associated with a 5-yearRFS of 14%. However, high clinical risk score does notpreclude 10-year survival in that patients with high scoresstill demonstrate actual 10-year survival of up to 16%.1

Furthermore, with the exception of positive liver resectionmargin, there is not a single clinical or pathologic factorthat precludes 10-year survival (Table 1). These types ofscoring systems serve as a general guide for prognosis butare not ideal in that they do not specifically impacttreatment decisions and do not predict universally good oruniversally bad outcomes.

Not surprisingly, some of the same clinical factors (i.e.,node-positive primary tumor) are predictive of outcome atother institutions. Despite this, these scoring systems arenot always prognostic across institutions. The MayoClinic devised a scoring system from their patient cohortthat included factors such as positive hepatoduodenallymph nodes, perioperative blood transfusion, node-positive primary tumor, disease-free interval, and sizeand number of metastatic lesions.5 While validating theirscoring system, they imported the data from their cohortinto three other scoring systems, including the MemorialCRS.5 Survival and recurrence were not stratified by anyof the scoring systems from other institutions (concor-dance indexes for all systems approximated 0.55). Clearly,there is significant variability which highlights thecomplexity of developing a reliable and reproduciblescoring system independent of surgeon and institutionalbiases.

There are numerous possibilities for why scoring systemsare not generalizable, and one of them is clearly an overallselection bias. Surgeons are typically very good at choosingappropriate surgical candidates who have demonstrated gooddisease biology, which limits the ability to generalize topatients with very high or very low clinical risk scores. Thereis also variability in the selection bias in that at differentinstitutions patient selection, referral patterns, and institutional

Years after resection

Ove

rall

Surv

ival

(%)

5 10

50

100

A

B

0

B

A

Months after randomization

Ove

rall

Surv

ival

(%)

Wild type k-ras

Mutated k-ras

Fig. 1 Prognostic markers in cancer and disease. a Theoretical exampleof the ideal biomarker. The ultimate biomarker is a predictor ofoutcome that is simple, cheap, easy to measure, and predicts majordifferences in outcome in patients with a given disease. b Overallsurvival of patients with advanced colorectal cancer based on treatmentwith cetuximab and K-ras status. Patients with wild type K-ras derived asignificant benefit from cetuximab while those with mutated K-ras hadequivalent outcome. Adapted from Karapetis et al.6 with permission

J Gastrointest Surg (2011) 15:406409 407

neoadjuvant and adjuvant paradigms differ. Another factor istumor biology, which ultimately dictates outcome and ispoorly understood. Taken together, selection bias and a poorunderstanding of tumor biology leads to the development ofrisk scores that are not generalizable and that do not impacttreatment decisions. Overall, there is no scoring system to datethat fits the paradigm demonstrated in Fig. 1 (the idealprognosticator).

Nomograms have been increasingly developed andutilized as prognosticators in multiple malignancies. Anomogram for predicting disease-specific survival afterresection of CRLM was recently published from ourinstitution.7 Ten factors were weighted and scored from ananalysis of 1,477 patients treated from 1986 to 1999(Fig. 2). The concordance index was 0.68 compared to

0.65 for the CRS. This nomogram has been recentlyvalidated at another institution and was found to be morepredictive of outcome than the CRS.8 Despite thesepositive findings, nomograms suffer from the sameinherent biases as other scoring systems and are notproven to be clinically helpful in guiding treatmentdecisions.

Neoadjuvant chemotherapy is being utilized withincreasing frequency and response to therapy is anotherpotential biomarker of patients with CRLM. However, itis clear from prospective phase II and III trials that therate of progression on neoadjuvant chemotherapy is lessthan 10%. Older studies suggest a poor outcome forpatients that progress on systemic chemotherapy. How-ever, newer studies in the era of modern chemotherapy

Table 1 Clinicopathologic variables from 612 patients treated at MSKCC with 10 year follow-up and impact on overall survival

Variable 10-Year survival (%)

Synchronous disease 13 11 5 7

Node-positive primary tumor 63 56 52 50

Preoperative CEA >200 ng/mL 16 11 8 7

Disease-free interval 1 59 51 32 39

Size of hepatic metastases >5 cm 53 41 41 35

Positive resection margin (liver) 20 10 9 0

Resection > or = lobectomy 63 63 62 68

> or = 4 hepatic metastases 23 16 11 5

Adapted from Tomlinson et al.1 with permission

CEA carcinoembryonic antigen

Fig. 2 Nomogram for predict-ing 96 months disease-specificsurvival. Draw a straight line foreach patient variable up to thepoint axis. The cumulativenumber of points correlates tothe disease-specific survivalprobability for a given patient.Taken from Kattan et al.7 withpermission

408 J Gastrointest Surg (2011) 15:406409

have shown that progression during neoadjuvant chemo-therapy does not necessarily preclude a good outcomeafter resection.9 Therefore, response to neoadjuvantchemotherapy alone is of low yield as a predictive markerand is not a reliable prognosticator independent of otherclinical risk factors.

It is clear that we need something better that, in the era ofmodern chemotherapy, can help guide treatment decisions.The hope of an ideal prognostic marker will likely have tocome from the benchtop where tumor biology can bepredicted independent of the aforementioned selection biases.The answers likely reside in tissue and serum banks which,with new technologies and better understanding of tumorgenetics, are amenable to future study. Tumor immunologicand inflammatory response, markers of sensitivity to certainchemotherapies (i.e., thymidylate synthetase), chemokines,and tissue microarray profiling have all demonstrated theability to prognosticate tumor biology.10 However, thesestudies have been small in number and at single institutionsand need further validation.

In conclusion, better prognostic factors are needed toguide the treatment of patients with resectable CRLM.Despite the numerous staging and scoring systems that existto stratify outcomes, they have limited utility in that theyare not reproducible and do not define either universallygood or bad outcomes. To improve risk stratification, weneed to explore the prognostic factors related to tumorbiology that are independent of the currently utilizedclinical and pathologic variables.

References

1. Tomlinson JS, Jarnagin WR, DeMatteo RP, et al. Actual 10-yearsurvival after resection of colorectal liver metastases defines cure.J Clin Oncol 2007; 25(29):457580.

2. ReesM, Tekkis PP, Welsh FK, et al. Evaluation of long-term survivalafter hepatic resection for metastatic colorectal cancer: a multifacto-rial model of 929 patients. Ann Surg 2008; 247(1):12535.

3. Fong Y, Fortner J, Sun RL, et al. Clinical score for predictingrecurrence after hepatic resection for metastatic colorectal cancer:analysis of 1,001 consecutive cases. Ann Surg 1999; 230(3):30918; discussion 31821.

4. Iwatsuki S, Dvorchik I, Madariaga JR, et al. Hepatic resection formetastatic colorectal adenocarcinoma: a proposal of a prognosticscoring system. J Am Coll Surg 1999; 189(3):2919.

5. Zakaria S, Donohue JH, Que FG, et al. Hepatic resection forcolorectal metastases: value for risk scoring systems? Ann Surg2007; 246(2):18391.

6. Karapetis CS, Khambata-Ford S, Jonker DJ, et al. K-ras mutationsand benefit from cetuximab in advanced colorectal cancer. N EnglJ Med 2008; 359(17):175765.

7. Kattan MW, Gonen M, Jarnagin WR, et al. A nomogram forpredicting disease-specific survival after hepatic resection formetastatic colorectal cancer. Ann Surg 2008; 247(2):2827.

8. Reddy SK, Kattan MW, Yu C, et al. Evaluation of peri-operativechemotherapy using a prognostic nomogram for survival afterresection of colorectal liver metastases. HPB (Oxford) 2009; 11(7):5929.

9. Gallagher DJ, Zheng JT, Capanu M, et al. Response to Neo-adjuvant Chemotherapy Does Not Predict Overall Survival forPatients With Synchronous Colorectal Hepatic Metastases. Annalsof Surgical Oncology 2009; 16(7):18441851.

10. Katz SC, Pillarisetty V, Bamboat ZM, et al. T cell infiltratepredicts long-term survival following resection of colorectalcancer liver metastases. Ann Surg Oncol 2009; 16(9):252430.

J Gastrointest Surg (2011) 15:406409 409

SSAT/AHPBA JOINT SYMPOSIUM 2010

The Role of Peri-operative Chemotherapyfor Resectable Colorectal Liver Metastasis: What Doesthe Evidence Support?

Timothy M. Pawlik & David Cosgrove

Received: 2 January 2011 /Accepted: 11 January 2011 /Published online: 21 January 2011# 2011 The Society for Surgery of the Alimentary Tract

Introduction

With improved patient selection, better surgical techniques,and more effective cytotoxic chemotherapy agents, 5-yearsurvival following curative intent surgery of colorectalmetastasis now approaches 4560%.14 While there havebeen significant advances in prolonging overall survival ofpatients with colorectal liver metastasis, many patients stilldevelop recurrent disease. De Jong et al.5 reported acontemporary experience in which the 5-year disease-freesurvival was only 30% following curative intent surgery forcolorectal liver metastasis, with 60% of patients developingextrahepatic disease at 5 years. Tomlinson et al.6 noted thatapproximately one third of actual 5-year survivors succumbto cancer-related death. Noting that the chance of cureafter hepatectomy was roughly a one-in-six chance, theauthors estimated a maximal cure rate of only about 25%for patients undergoing surgical resection of colorectal livermetastasis. Given the persistent high recurrence rates andthe overall poor true long-term survival followingsurgical resection of colorectal liver metastasis, there has

been great interest in the use of adjuvant chemotherapy forpatients with resectable colorectal liver metastasis.

The role of adjuvant chemotherapy after resection ofcolorectal cancer liver metastasis has recently beenreviewed by Power and Kemeny.7 For both pedagogicaland practical purposes, peri-operative chemotherapy forcolorectal liver metastasis can be divided into threedifferent treatment strategies; neoadjuvant, peri-operative, and adjuvant. We herein review each one ofthese peri-operative chemotherapy treatment strategies forresectable colorectal liver metastasis.

Adjuvant Chemotherapy

The use of adjuvant chemotherapy for colorectal cancer hasbeen well studied in stage III disease. Specifically, multiplerandomized clinical trials have noted a survival benefitassociated with the use of adjuvant chemotherapy forpatients with colorectal cancer and lymph node metasta-sis.811 Sargent et al.8 reported that surgery plus adjuvant 5-flurouracil (5-FU) versus surgery alone was associated withan overall survival benefit (8-year overall survivalsurgery+5-FU-based chemotherapy, 53% versus surgeryalone, 43%; P

adjuvant chemotherapy in the management of stage IIIcolorectal cancer.

Multiple phase I and II studies have similarly shownimproved efficacy of modern-era chemotherapy in the treat-ment of unresectable stage IV colorectal liver metastasis. Whilemonotherapy with 5-FU previously resulted in response ratesonly in the range of 2025%,12 current regimens that includeoxaliplatin or irinotecan have response rates in the range of4555%.1317 More effective cytotoxic chemotherapy hastranslated into a significant increase in the median survival ofpatients with unresectable colorectal liver metastasis from6 months with best-supportive care to 1215 months withmonotherapy 5-FU to now 2024 months with oxaliplatin- oririnotecan-based therapies. Additional advances have beenassociated with the addition of biologic agents, such asbevacizumab or cetuximab, to cytotoxic chemotherapy, asoutlined in the BEAT18 and CRYSTAL studies.19

Given the robust data on the role of systemic chemo-therapy for both resected stage III and unresectable stage IVcolorectal cancer, there has been interest in the potential useof adjuvant chemotherapy in the setting of resected stage IVcolorectal liver metastasis. Unfortunately, data on the roleof adjuvant chemotherapy for resected colorectal livermetastasis are scant.2023 Of the four randomized trialspublished to date, two were published only in abstract formand each had fewer than 52 patients analyzed.20,22 Of thetwo other randomized trials,21,23 both suffered from pooraccrual and had fewer than 175 patients analyzed. In theLanger et al.21 study, only 107 patients were analyzed andthere was no noted difference in overall survival when dailybolus 5-FU was compared with observation alone followingresection of colorectal liver metastasis. In the Portier et al.23

trial, 173 patients were randomized to surgery alone versussurgery+5-FU. In this study, two patients were lost tofollow-up, leaving 85 patients randomized to the surgeryalone arm versus 86 patients to the surgery+5-FU arm.Among the patients randomized to adjuvant 5-FU, 94% ofassigned patients received post-operative chemotherapy. Nodifference in overall survival was noted between the studyarms (5-year overall survivalsurgery alone, 42% versussurgery+5-FU, 51%; P=0.13). The authors did note,however, that adjuvant 5-FU conferred a disease-freesurvival benefit (5-year disease-free survivalsurgeryalone, 27% versus surgery+5-FU, 34%; P=0.028). Coxmultivariate analysis confirmed a statistically significantbeneficial effect of chemotherapy on disease-free survival(HR=0.66, 95% CI, 0.460.96).23 While this study didshow an improvement in disease-free survival, it failed toshow an overall survival benefit with adjuvant therapy. Thereasons for this lack of effect were undoubtedly multi-factorial and included the relatively small study sample sizeand lack of statistical power. In an attempt to increase theoverall number of patients available for analysis, Mitry et

al.24 performed a pooled analysis of the Langer et al.21 andPortier et al.23 studies. In this pooled analysis, a total of 278patients were analyzed, 140 of whom had been randomizedto surgery alone and 138 of whom had been randomized tosurgery+5-FU. Among those patients randomized toadjuvant chemotherapy, 95% of assigned patients receivedchemotherapy. In this study, the authors again noted nodifference in overall survival and a trend towardimproved disease-free survival associated with adjuvanttherapy. On multivariate analysis, after controlling forother competing risk factors, a marginal statisticallysignificant associated benefit of adjuvant 5-FU chemo-therapy was noted (P=0.046).

There have been several large retrospective, non-randomized studies that have also examined the issue ofadjuvant chemotherapy for resectable colorectal livermetastasis.2527 Each of these studies have reported asurvival benefit for adjuvant 5-FU versus surgery alonefor patients with resectable colorectal liver metastasis. Ingeneral, these studies have noted a relative 2560%decreased risk of disease-specific death associated withreceipt of adjuvant 5-FU. Obviously, these retrospectivestudies have serious threats to validity including selectionbias and treatment bias, not to mention issues with possibleconfounding. As such, any causal inferences drawn fromsuch data need to be carefully considered.

There has been one study that has examined the use ofmodern era chemotherapy in the adjuvant setting forresected colorectal liver metastasis. Ychou et al.28 reported arandomized phase III study comparing adjuvant 5-FU versusFOLFIRI among patients having undergone complete resec-tion of liver metastases from colorectal cancer. In this study,321 patients were randomized to receive either 5-FU alone orFOLFIRI. Of those patients assigned to FOLFIRI, 95%received the assigned post-operative chemotherapy. Theauthors noted no benefit for FOLFIRI compared with 5-FUwith regards to either diseas