Thorax (1973), 28, 667.

Reconstruction of the tracheaExperience in 100 consecutive cases'

HERMES C. GRILLO

Massachusetts General Hospital, Boston, Massachusetts, USA

Grillo, H. C. (1973). Thorax, 28, 667-679. Reconstruction of the trachea. Experience in 100consecutive cases. Anatomic mobilization of the trachea permits resection of one-half ormore with primary anastomosis. An anterior approach by a cervical or cervicomediastinalroute utilizes cervical flexion to devolve the larynx and tracheal mobilization with preservationof the lateral blood supply. The transthoracic route is employed for lower tracheal lesions.Over 100 tracheal resections have been done using these methods of direct reconstruction.

Eighty-four patients suffered from benign strictures, 79 resulting from intubation injuries.Eleven primary tracheal tumours and five secondary tumours are included. The majority oflesions following intubation occurred at the level of the cuff. It was possible to repair 78 ofthe 84 stenotic lesions through a cervical or cervicomediastinal approach. Seventy-three ofthe 84 patients with inflammatory lesions obtained an excellent or good functional andanatomic result. Nine of 11 patients with primary neoplasms who underwent reconstructionare alive and without known disease. There were five early postoperative deaths in these100 consecutive patients who underwent tracheal reconstruction.

PROBLEMS OF RECONSTRUCTION

Until a decade ago reconstruction of the tracheawas inhibited by the belief that no more than2 cm of trachea-approximately four cartilaginousrings-could be circumferentially resected and thetracheal ends predictably reapproximated (Belsey,1950). Faced with a more extensive lesion thesurgeon felt, therefore, that he must abandon theideal of end-to-end anastomosis of the trachea infavour of endoscopic removal -of the lesion, lateralresection with patching of the defect by one ofa variety of means, or insertion of a tubularprosthesis. Endoscopic removal was obviously onlya temporizing manoeuvre to provide an airway inthe case either of a neoplastic lesion or of mostbenign strictures. Lateral resection, performedwith concern for maintaining a good isthmus ofnormal tracheal wall, all too often compromisedresection margins and led to a recurrence oftumour. Problems with the integrity of lateralpatches also arose, depending on the material used.A number of case reports of at least initial success

lBased upon a paper given in September 1972 before the Societyof Thoracic and Cardiovascular Surgeons of Great Britain andIreland at the University of Sheffield. Experimental work describedwas supported in part by a research grant from the John A. HartfordFoundation, Inc.

Requests for reprints: Massachusetts General Hospital, Boston,Massachusetts 02114 USA.

with prostheses have been recorded. No prosthesishas yet been developed, however, which may beinserted as a replacement for long segments oftrachea with a high predictability of success(Grillo, 1970). Although early leakage with resul-tant mediastinal sepsis may be reduced by sealingwith homologous tissues, the problems of latestenosis at either end of the prosthesis and erosionof major mediastinal vessels have continued toplague the tracheal prosthesis. Placement of aforeign material, however inert, in a bed of con-nective tissue contiguous with potentially infectedrespiratory mucosa remains a biologically unstablesituation. Indeed there are no examples of uniformsuccess of other non-tracheal prostheses inanalogous situations. The presently useful pros-theses-cardiac, vascular, orthopaedic-are allseated entirely in potentially sterile mesenchymaltissues. The somewhat lesser obstacles of anaes-thetic management of patients during airwayreplacement also seemed to pose dilemmas adecade ago.

ANATOMIC MOBILIZATION

In view of the then and as yet unsolved problemsof providing a suitable biological basis for the useof prostheses, I began to reinvestigate systemati-cally the problem of anatomically mobilizing the

667

Hermes C. Grillo

human trachea to permit end-to-end reconstruc-tion following the resection of greater segmentsof trachea. This work was prompted by the obser-vation in cases of traumatic severance of thetrachea that the distal end of the divided cervicaltrachea dropped into the mediastinum. Theseanatomic studies (Grillo, Dignan, and Miura,1964) were done in the pathology laboratory,since no close parallel existed in gross relationshipsbetween human and other animal tracheas. Sincemy viewpoint was that of a thoracic surgeon, myinitial attempts were to mobilize the thoracictrachea. Our first observations demonstrated thatdivision of the inferior pulmonary ligament andmobilization of the right hilus with loosening ofcarinal attachments permitted resection of an

average of 3 cm of trachea or six rings. Recon-struction was possible at tensions well under1,000 g. Pericardial dissection of the pulmonaryvessels allowed the additional removal of 0 9 cm.

V/ I

4.5

1.4

I)

FIG. 1. Diagram of lengths of trachea which may beresected and yet permit primary anastomosis. (a) 4.5 cm

can be removed through an anterior approach and approxi-mation done with 1,000 g of tension, utilizing cervicalflexion. Intrathoracic dissection allowedfurther removal of1-4 cm. (b) With the cervical spine in a neutral position,transthoracic hilar dissection and division ofthe pulmonaryligament allowed 3 cm to be removedand anastomosis done.Intrapericardial dissection added09 cm and division of theleft main bronchus with reimplantation of the bronchusintermedius permitted an additional 2-7 cm to be removed.Ifcervicalflexion is utilized, a significantly greater amountthan 3 cm may be removed.

A large increment (2-7 cm) could be gained bydividing the left main bronchus and reimplantingit into the bronchus intermedius. Thus, removalof approximately one-half of the trachea and itsreconstruction appeared to be possible (Fig. la).Initial studies were done with the patient's neckin a neutral position and so later proved to be aconservative statement. As our anatomical studiesprogressed it became clear that cervical flexionalone would permit devolvement of cervicaltrachea into the mediastinum so that extensivesegments could be removed without having toresort to the additional hazard of reimplantationof the left main bronchus. This last manoeuvrehas been reserved by me in clinical practicefor carinal reconstruction (Grillo, Bendixen,and Gephart, 1963). The technique was first usedclinically by Barclay, McSwan, and Welsh (1957).Experimental measurements of the tensionsdeveloped following reanastomosis after removalof successive centimetres of trachea indicated thatthese tensions would lie within what appearedto be experimentally safe levels. Cantrell and Folse(1961) observed that tracheal separation did notoccur when the stress was under 1,700 g. In bio-logical measurements such as this, there is everyreason to accept observations from animal workas applicable to man.

In our early experiments we also proposed thepossibility of devolving a segment of cervicaltrachea directly into the mediastinum, reconstruc-tion to be done later in the neck. We (Griulo,Dignan, and Miura, 1966) then developed amethod for construction of a full thickness skintube supported by totally buried prostheticmaterial for either this eventuality or for thereplacement of cervical trachea. However, after itsuse in a few clinical cases (Grillo, 1965) furtheranatomical investigation made the procedureobsolete for most situations. In further anatomicalstudies (Mulliken and Grillo, 1968) we sought todetermine how much of the cervical and uppermediastinal trachea could be removed by ananterior approach and reconstruction effected.We utilized cervical flexion to a 'comfortable'position of not more than 350, placed tension of1,000 g on the trachea, and found that 4*5 cmmight be removed. The addition of intrathoracicdissection added another 14 cm of potential length(Fig. lb). In practice, we have been able to removeone-half or more of the trachea in a number ofpatients through the upper approach, utilizing onlycervical flexion and such mobilization of the distaltrachea as may be obtained from the anteriorapproach. Dedo and Fishman (1970) havedescribed a technique for laryngeal devolvement

668

I'I

Reconstruction of the trachea

which adds a further 1-5 cm of anterior mobility.This technique, in conjunction with anatomicalmethods described, usually obviates the need forintrathoracic entry in most lesions. It should notbe used as a primary manoeuvre.

FIG. 2. The blood supply ofthe tipper trachea is principallyfromfine branches ofthe inferior thyroidartery. This bariumsulphate arterial injection demonstrates the most common

pattern.

A critical factor in such mobilization techniquesis preservation of the relatively tenuous trachealblood supply. The lower trachea shares its supplywith the oesophagus from the bronchial arteries.Precise information was not available regardingthe supply of the upper trachea, and additionalstudies were necessary (Miura and Grillo, 1966).In general, the upper trachea is supplied by threebranches from the inferior thyroid artery, andthere is only the finest of collateral pathwaysbetween these branches (Fig. 2). Each branchsupplies the trachea anteriorly and sends a divisionposteriorly to the oesophagus. The vessels are ofsmall diameter. In mobilization of the tracheacare is therefore taken to free the tracheaanteriorly and posteriorly, leaving the lateralpedicles intact except for a short distance whichis to be used in the anastomosis. The absence ofa pedicled blood supply may be of considerablenegative importance in the future if transplanta-tion of organs becomes more widely practicable.

TABLE ITRACHEAL RECONSTRUCTION (SINCE 1962)

Cases Deaths

InflammatoryPost-intubation.79 3Post-traumatic.4 0Idiopathic 0I

NeoplasticPrimary.11 1Secondary 5 1

Total.100 5

CLINICAL APPLICATIONS

These anatomic methods of tracheal mobilizationfor reconstruction have been applied in the lastdecade to over 100 patients (Table I). Eighty-fourof these patients suffered from benign inflamma-tory lesions. Of these, 79 were the consequenceof intubation, usually for ventilatory support, fourfollowed direct tracheal trauma, and one was ofidiopathic origin. Eleven primary tracheal tumoursand five secondary tumours were managed byexcision and direct reconstruction. There were fivedeaths in this series of reconstructions. Two ofthese fatalities may have been preventable in thelight of the experience gained over 10 years. Thisreport is based on the lessons learned in the accom-plishment of these 100 reconstructions. Duringthe same period a marked number of patientswith tracheal diseases were seen who did notundergo surgical treatment, either because theirbasic medical disease precluded surgical correctionof the tracheal lesion or, in a small number,because of the extent of the tracheal disease itself.In addition, a number of patients underwenttracheal surgery other than end-to-end reconstruc-tion. These include patients who required resectionof the larynx as well as the trachea for neoplasmor underwent corrective procedures such as plasticclosure of persistent tracheal stomas or surgicalsplinting for tracheomalacia.My initial interest in extending the range of

feasibility in surgery of the trachea was to improvethe management of those rare primary tumoursof the trachea which presented. A retrospectiveview of the Massachusetts General Hospital'sexperience with tracheal tumours suggested thatmany were of low malignancy. This was indicatedby the loag course of symptoms prior to theirdiagnosis and the fact that the tumours oftenappeared to be localized even at the time of thepatient's death. Since death was so often due toairway obstruction, potential curability or at leastlengthy palliation was suggested by these observa-tions. Most primary tumours of the trachea, ex-cluding carcinoma of the larynx extending into

669

Hermes C. Grillo

the trachea and those of the lung or oesophagusinvading the trachea, are nonetheless epidermoidin type (Fig. 3). In approximately 30 years, 55patients with tracheal tumours were seen at theMassachusetts General Hospital (Grillo, 1 972b).Thirty of these were epidermoid carcinoma, 19adenoid cystic carcinoma (formerly called cylin-droma), and seven were of varied histology. It isinteresting to note that in a similar period of timetwo other major institutions had a very similarexperience. The Mayo Clinic (Houston, Payne,Harrison, and Olsen, 1969) reported 53 cases, ofwhich 24 were epidermoid, 19 adenoid cystic, and10 of mixed variety. Memorial Hospital for Cancerand Allied Diseases (Hajdu et al., 1970) counted41 tumours, of which 30 were epidermoid andseven adenoid cystic. Secondary tumours, whichmost frequently involve the trachea, in addition tothe larynx, lung, and oesophagus, are those of thethyroid, other head and neck lesions, breast, andlymphoma. I believe that involvement of thetrachea by secondary tumour usually indicates anadvanced stage of an aggressive lesion whichcontraindicates extension of surgical proceduresto include segments of trachea. On the other hand,there are a number of thyroid lesions which arebetter treated by concomitant or even late trachealresection, if only to provide long-term palliation.I have included tracheal resection in some resec-tions for carcinoma of the oesophagus where com-bined radiation therapy followed by surgery havebeen employed and where the only point of

AA\.1

I i..I I_lk

FIG. 3. Primary tracheal neoplasms. (a) Squamous-cellcarcinoma of trachea removed segmentally in 1964 bytransthoracic route. Patient remains free of disease.(b) Squamous-cell carcinoma of upper trachea removed byan anterior approach. One node was positive and there was

a paratracheal recurrence three years later. (c) Carcinoidadenoma obstructing lowermost trachea just above thecarina.

FIG. 4. Diagram of principal lesions resulting from theuse of cuffed tracheostomy tubes. (A) Conventionaltracheostomy tube andhigh-pressure cuff in situ. (B) Lesionsoccur at several levels. At the stoma (a) anterolateralstenosis may occur, a granuloma (b) or a combination ofboth. At the cuff level (c) circumferential stenosis results.Varying degrees of malacia (d) may occur in the segmentbetween the stomal and cuff levels. In children, particularly,granulomas may occur anteriorly where a tip of anuncuffed tube has caused ulceration of the tracheal wall (e).

670

Reconstruction of the trachea

° cm

FIG. 5. Stenotic lesions resulting from tracheos-tomy tubes. (a) Stomal stenosis. Anterolateral stric-ture results in a triangular shaped lumen with theapex anteriorly. In this patient the back wall of thetrachea has sustained some damage and showsgranulation tissue. Often the back wall is uninjured.(b) Cuff stenosis. The original damage was circum-ferential and the subsequent stenosis is thereforecircumferential.

apparent non-resectability of the lesion has beenfixation to the trachea.

Coincident with our initial development of an

anatomical approach to tracheal reconstruction,inflammatory lesions of the trachea secondary tosuccessful ventilator therapy began to appear inincreasing numbers. Lesions in this category con-

fused the clinician. However, following intensivestudy of pathological specimens of the tracheas ofpatients who failed to survive ventilator therapy(Cooper and Grillo, 1969a; Florange, Muller, andForster, 1965) and growing experience of surgicalremoval of these lesions (Grillo, 1969a) and,finally, after prospective analysis of trachealdamage by various types of intubation equipment(Andrews and Pearson, 1971) necrosis becameindicted as the primary cause of a variety oflesions. The two principal areas of stenosis werethose at the level of the stoma and at the level ofthe cuff (Fig. 4). Lesions at stomal level wereanterolateral because the original damage hadbeen anterolateral, and those at cuff level werecircumferential consequent upon the presence ofthe circular cuff (Fig. 5). In addition, a variety ofother lesions was noted, including tracheomalaciaboth at the cuff level and in the intervening seg-ment between the stoma and the cuff, granulomasat stomal level and at the level of the tube tip, andfull-thickness tracheal erosions creating tracheo-innominate fistulas and tracheo-oesophagealfistulas. These syndromes have now been preciselydescribed (Geffin, Grillo, Cooper, and Pontop-pidan, 1971; Grillo, 1970; Harley, 1971) and thediagnostic studies necessary for preoperativedelineation have also been detailed (Macmillan,James, Stitik, and Grillo, 1971). In recent workwe have reproduced these lesions experimentally(Cooper and Grillo, 1969b) and have devised lowpressure cuffs which have the potential to eliminatecuff stenosis (Griuo, Cooper, Geffin, and Pontop-pidan, 1971). The incidence of stomal striotureshas also been reduced markedly by the use oflight-weight connectors to avoid erosive leverageat the stomal level.

This report is not primarily concerned with thisbroad field of iatrogenic disease but the informa-tion is necessary background for the considerationof resectional treatment of these lesions. In oursurgical patients treated by resection and anasto-mosis 81 had post-intubation stenoses of thetrachea (Table IL). Fifty-three of these lesions wereat cuff level and 22 at stomal level. In five patientsboth lesions were present and in one it was notpossible to tell what the origin had been. In addi-tion, four -patients with post-traumatic stenosis ofthe trachea and one with an idiopathic inflam-matory stricture have been corrected surgically.Twenty-one of the patients had their originalventilatory treatment at the Massachusetts GeneralHospital. Sixty of these were referred from otherinstitutions. Five patients had undergone previousresection elsewhere prior to referral but hadsuffered prompt recurrence of the stricture.

f I I

671

2

Hermes C. Grillo

TABLE IITRACHEAL STENOSIS (SURGICAL CASES)

Post-intubation 811Stoma 22Cuff 53Both 5Uncertain 1

Post-traumatic 4Idiopathic 1

1MGH 21Referred from other hospitals 60.

OPERATIVE APPROACH

The clinical common denominator of 'surgical'lesions of the trachea is most often airway obstruc-tion whether from neoplasm or inflammation. Thesurgical problem is largely that of relief fromobstruction and anatomical reconstruction of theairway so that it will provide normal function.Approach must obviously vary with the type andextent of pathology (Griulo, 1969b; 1972a).Detailed radiographs are obtained, except in emer-gency circumstances, to define the exact level ofthe lesion, its extent, and severity, as well as toobtain information about laryngeal function andthe presence or absence of attendant malacia.We have not found contrast media to be par-ticularly helpful. Any tracheostomy tube presentis removed during these examinations to obtainmaximum information about the lesion. A physi-cian must always be present to reinsert such atube since many patients will close their airwaysdown rapidly and alarmingly during the courseof the radiological examination. Bronchoscopy isusually deferred until the time of surgery to avoidthe possibility of precipitating acute obstructionfollowing endoscopy. Anaesthesia will not bedescribed in detail here (Geffin et al., 1969). It isdesigned to provide a slow, gentle induction with-out precipitating obstruction, to allow the patientto breathe spontaneously throughout the case, withassistance where necessary, so that he may beextubated promptly at the conclusion of the pro-cedure and the airway evaluated prior to hisleaving the operating room. If the degree ofobstruction is not very great, an endotracheal tubeis placed above the lesion, and through thisadequate ventilation is obtained. In cases ofsevere stenosis (airway less than 4 mm in diameter),dilatation is performed just prior to exploration,using serial paediatric ventilating bronchoscopes,and a small endotracheal tube is passed throughthe stenosis.

Benign lesions of the upper trachea areapproached initially through a collar incision withthe addition of a vertical limb to divide the upperhalf of the sternum, if necessary (Fig. 6a). In

particularly extensive lesions the patient is posi-tioned on the table so that the incision may becontinued into the right fourth interspace to pro-vide access to the thoracic trachea. This type ofincision is the only one which provides access tothe entire trachea from cricoid to carina. Suchpositioning is advisable with a neoplastic lesionof the upper trachea. An existing stoma is handledvariably. It is either included or not with theoriginal collar incision. The incision, however, isalways placed low and, if the stoma is not in theline of incision, it is separately excised if necessary.Occasionally, if a stricture lies far enough belowan existing stoma, the opening may be left un-disturbed. If, however, with the shift in position

(a)"S

_(d..)

16. 6.

672

(tr)

Reconstruiction of the trachea

carefully dissects the nerves, isolating theminitially at a distance from the lesion. Unfortun-ately, extensive dissection of all paratracheal tissueon both sides cannot be done for the full lengthof the trachea without running into the danger ofdevitalizing the proximal end of the distal segmentof trachea which is to be left in place.

If there is insufficient space in which to workor in which to isolate the lesion because of thepatient's particular anatomy, the upper sternumis divided and exposure is obtained to a pointbelow the angle of Louis with a small chest

(c)

of the larynx and trachea the skin pulls too tightlyon its stomal tether, the skin is detached and thestoma is exteriorized at an appropriate level. Onrare occasions when such a stoma is devolvedbelow the sternum in the course of reconstruction,it must beF closed with a muscle flap. If, however,the stoma and its margins are too close to acircumferential cuff lesion, the resection mustinclude both lesions so that the suture line willnot lie in inflamed tissue. Dissection initiallyexposes the entire anterior surface of the trachea,in most cases, from cricoid cartilage to carina.When there is dense adherence of inflammatorytissue in benign strictures, dissection is kept hardon the trachea to avoid injury to the recurrentnerves. In particular, when dealing with a highsubcricoid stenosis it is necessary to expose theinferior margin of the cricoid circumferentiallywithout, however, dissecting up laterally farenough to injure the entry point of the recurrentnerves into the cricothyroid membrane. Theapproach for neoplasm is quite different, in thatthe operator attempts to take as much trachealtissue in the region of the lesion as is possible and

(d)

FIG. 6. Upper tracheal reconstruction: (a) The collarincisionpermits exploration and, in many cases, reconstruc-tion. Upper sternal division allows access to the medias-tinumposterior to the great vessels. (b) The innominate veinis retracted since its division will not permit greater expo-sure. Thepleura remains intact. (c) Following division ofthetrachea below the lesion, intubation is carried out across theoperative field. The adherent inflammatory lesion is dis-sectedfrom the oesophagus. (d) Details ofsuture technique.After the posterior two-thirds of the sutures have beenplaced, the tube above may be advanced distally or allsutures may beplacedprior to removing the tube across thefield.

673

Hermes C. Grillo

retractor (Fig. 6b). It is not necessary to divideacross the sternum into an interspace nor is itnecessary to carry the incision down to the bottomof the sternum. The trachea quickly becomes aposterior mediastinal organ and the principal con-tribution of the sternal division is to allow work-ing space at the thoracic inlet. Similarly, divisionof the innominate vein contributes nothing sincethe innominate artery very shortly presents itselflying anterior to the trachea. Care is taken not toenter the pleura, but if this occurs a chest tubeshould be placed or at least the pleura evacuatedcompletely before closure. The anterior approachis particularly well tolerated by patients with poorlung function.

After exposing the anterior surface of the trachea,dissection is carried circumferentially around thetrachea, usually at a point immediately below thelesion so that not too much trachea will be cir-cumferentially freed of its blood supply. Afterthis, mid-lateral traction of sutures of 2-0 silkare placed through the full thickness of tracheaapproximately 2 cm below the level of potentialtransection of the trachea. Similar traction suturesare applied above in either the upper trachea orthe larynx, as the level of resection requires.Following this, the trachea is transected belowthe lesion, and as much normal trachea as possibleis saved. If necessary, a tentative opening is madein the case of a benign lesion and the interior ofthe trachea is examined before the final level oftransection is selected. Intubation may now beperformed across the operative field using pre-selected flexible Tovell endotracheal tubes andsterile anaesthesia tubing, the ends of which arepassed to the anaesthetist (Fig. 6c). A cuffed tubeis used to minimize the amount of blood whichenters the tracheobronchial tree. In addition, con-tinuous attention is paid to suctioning blood whichpuddles above the cuff. This minimizes early post-operative shunting. The proximal lesion is now

elevated to facilitate resection of the lesion fromthe adjacent oesophagus to which it may be denselyadherent if inflammatory, or which it may haveinvaded if neoplastic. The proximal and distaltrachea are also partially freed from the oeso-phagus posteriorly so that lateral pedicles aredeveloped. The neck is then tentatively flexedand traction is made on either side on the upperand lower traction sutures to see if approximationmay be easily obtained. If there appears to be anunsatisfactory amount of tension, additional bluntdissection at the carinal level may be accomplished.A second option, if the tension appears to be toogreat, is to raise the upper skin flap higher byextending and curving the collar incision upward

at either end in order to expose the larynx, andto perform a laryngeal release procedure. On rareoccasions only is it necessary to enter the thoraxthrough the right fourth interspace for mobiliza-tion of the hilus. This is most often necessary inthe case of upper tracheal tumours.

After demonstrating that approximation is pos-sible, the neck is re-extended and interrupted non-absorbable fine sutures are placed so that theknots will lie outside the trachea (Fig. 6d). I havemost recently been using 4-0 Tevdek with smallcutting needles. The sutures pass through the fullthickness of the tracheal wall and pass throughthe cartilage in the anterolateral two-thirds. Thesutures are carefully clipped to the field drapes sothat when the neck is flexed and the exposurediminishes, the sutures will be precisely locatedand may be tied sequentially from front to back.When the anastomosis is made to the bottom ofthe larynx, the sutures are similarly placed andgreat care is taken to make the anastomosis aslittle discrepant as possible. The endotracheal tubefrom above is next advanced after removal of thetube which lies across the field. The patient isplaced in full cervical flexion and propped up inthat position and the sutures are tied serially.Traction sutures have usually been tied first toprovide a tentative approximation but not intus-susception. The traction sutures are removed atthe conclusion of the anastomosis and its air-tightness is determined under saline. There shouldbe no leak. Postoperative tracheostomy tubes areavoided completely either above or below theanastomosis, and especially through it. On therare occasion when ventilatory support was un-expectedly required after reconstruction, a smallendotracheal tube has been left in place past theanastomotic line. Pressure of even a low pressurecuff against the anastomosis is highly undesirableand may lead to dehiscence and death. Afterwiring the sternum together and closing the inci-sions, with suction catheter wound drainage, aguardian stitch is placed just underneath the chinto the presternal skin to remind the patient thathe must keep his neck flexed. After a week thesuture is removed, but the patient is encouragedto keep his head in that position for at leastanother week to allow healing without tension onthe forming anastomotic scar.Approach to a lesion of the lower third of the

trachea is usually made through a high postero-lateral thoracotomy through the fourth interspaceor bed of a resected fourth rib (Fig. 7a-c). I usethis approach for tumours of the lower half ofthe trachea and carina. Only rarely should abenign stricture be approached through this route.

674

Reconstruction of the trachea

\i;\

(I'>*.

I<.

/I

(a

(c)FIG. 7. Reconstruction of the lower trachea: (a) Exposureis through a fourth interspace incision or the bed of thefourth rib. Full mobilization is accomplished as necessary.In the case of tumour, paratracheal nodal tissue is takenwith the specimen. If the patient's arterial oxygen fallssignificantly, the pulmonary artery may temporarily begently clamped while only the left lung is ventilated as in b.(b) The details of suturing are similar to that in the uppertracheal approach. The lateral traction sutures are usefulin initial approximation but are later removed. (c) A pleuralflap is placed over the anastomosis. Cervical flexion isutilizedfor approximation of the trachea in the transthora-cic approach also.

Although it may seem more attractive to thethoracic surgeon not versed in the anteriorapproach, it is much more traumatic to the patientand does not provide much advantage technically.Indeed it may make a simple resection difficult.The mobilization which is anticipated to be neces-sary is accomplished, if possible, before the divi-sion of the trachea. It is important for the anaes-thetist to flex the neck in anatomic relation tothe sternum to devolve as much cervical tracheaas possible into the mediastinum to minimize theamount of dissection. In cases of unusually exten-sive tumour I also prepare the cervical region andanterior chest for upward extension or forapproach into the neck for tracheal devolvementif necessary. Carinal reconstructions vary depend-ing on the exact location of the tumour. In general,our approach has been to remove the tumour(b)

675

Hermes C. Grillo

with an adequate margin and then improvise anappropriate reconstruction. We have found nouse for cardiopulmonary bypass in this kind ofsurgery. In the relatively simple cases it adds anunnecessary complication. In the most complexcases where enormous amounts of dissection andmanipulation of the lungs must be carried out,the danger of postoperative haemorrhage into themanipulated lung is too great to permit its safeuse.There is a group of patient's in whom even

partial salvage of the larynx is not possible dueto extension of tumour from the trachea. Forthese we have done extirpative laryngotrache-ectomy, removing thyroid or additional cervicalorgans as required (Fig. 8). Many hazards haveattended these destructive resections with thefashioning of low mediastinal tracheostomy. Pre-vious approaches which have attempted to pullthe trachea out to the surface have resulted intoo much tension. The development of compli-cated skin flaps to create cutaneous tubes to reachthe tracheal stump have also led to wound sepsisand, occasionally, to death from haemorrhagefrom an exposed innominate artery. In an effortto avoid these problems I proposed dropping anuncomplicated bipedicled anterior skin flap downto the stump of trachea after removing the anteriorplaque of sternum, heads of clavicles, and the twoupper rib cartilages (Grillo, 1966). In this flapwith excellent blood supply a circular stoma wasfashioned which could be sutured in the simplestpossible line to the stump of trachea even at the

r.- 5 56

FIG. 8. Adenoid cystic carcinoma invading trachea andthyroid gland. The larynx was also involved proximally.Laryngotracheectomy with removal of the overlying strapmuscles and the anterior wall of the oesophagus wasrequired. Despite a mediastinal tracheostomy the patientis functionally the same as the patient who has undergonelaryngectomy alone.

supracarinal level. Tension, devascularization, andcomplexity of suture line were thus avoided. None-theless, after initial successes, there have beeninstances of haemorrhage from the innominateartery (which lies immediately under the tracheo-cutaneous anastomosis) if primary healing did notfollow. Transplantation of this artery may beadvisable in the few cases where such destructiveresection is required.

In the group of 84 patients with tracheal stenosiswho underwent reconstruction (Table III), 50were accomplished with the cervical approachalone, 28 through a cervicomediastinal approach,five through a transthoracic approach, and in onea cutaneous tube was constructed. This representsan experience of eight years. Were the series beingdone again in the light of our experience, onlyone or two patients would have been approachedthrough the transthoracic route (for example,because of a right upper lobe tracheal bronchus).Only two of the patients with primary tumoursof the trachea were approached through theanterior cervicomediastinal route; their lesionswere of the upper trachea and were relativelyconfined, requiring no more than a 50% resectionof the trachea.

TABLE IIIRECONSTRUCTIVE TREATMENT FOR TRACHEAL

STENOSTS

End-to-end anastomosisCervical .. .. .. .. .. 50Cervicomediastinal .. . 28Transthoracic.5

Cutaneous tube.1

Total.84

RESULTS OF TREATMENT

The results of resection and end-to-end recon-struction have in general been quite satisfactory(Table IV). In the group of 84 patients withstenosis from all causes, 73 have an excellentfunctional airway and a good to excellentanatomic reconstruction judged radiologically.Four patients may be listed as satisfactory andable to accomplish all of their daily work withoutdifficulty, but two have varying degrees of post-operative narrowing at the anastomotic site onradiological examination. There were four failures,all of which could now be avoided. In one patientwe failed to recognize that a large segment of theanterior cricoid was destroyed and that there wastrue intralaryngeal damage. A second had a seg-ment of malacia in conjunction with stenosis.Because of poor pulmonary function the malaciawas not excised with the stenosis and later the

676

Reconstruction of the trachea

TABLE IVTRACHEAL STENOSIS: RESULTS OF RESECTION

Good, excellent 73Satisfactory . .. 4Poor, failures 4Deaths . .. 3

Total .84

patient required re-operation. A third patient hadsevere neurological damage and continued to haveaspiration problems thereafter which probablywere related more to the neurological problemsthan to the reconstruction which had been forcedby the anatomic situation. Three deaths occurredin the operative series. One desperately ill intu-bated patient had a recurrent double stenosisfollowing resection and insertion of a Marlexpatch elsewhere. The tracheal lesion was too closeto the carina to be managed by the endotrachealtube alone and reconstruction was attempted asthe only alternative. She required postoperativeventilation with the then available high-pressurecuff and this led to separation of the anastomosisin four days. A later patient developed bilateralpneumonia, had to be ventilated for two weekspostoperatively, and then also suffered separation.A third patient was salvaged from an acutetracheo-innominate artery fistula, had his innomi-nate artery resected, and underwent tracheal re-construction. Unfortunately, cardiac arrest duringthe acute incident led to irreversible neurologicaldamage.

In a group of 29 patients with primary trachealtumours who were seen during the period of thisstudy (1962-72) (Table V), approximately one-

TABLE VPRIMARY NEOPLASMS OF TRACHEA: MANAGEMENT

(1962-72)

WithoutCases Living Disease

Irradiation .. 10 2 0Complex resection .. 8 3 1Reconstruction . .. 11 9 9

Total .29

third (10 patients) had disease which was tooextensive for resection to be considered at thetime they were first seen. They received radiationor no treatment. At the present time only twoare living and both have known metastatic diseasein their lungs. Eight patients underwent varyingtypes of complex resection. Three of these patientsare living and only one is without disease. If we

focus on the 11 patients who underwent resectionwith intent to cure and with primary reconstruc-

3D

tion, nine patiernts are living and have been with-out disease for periods from six months to 10years. These include both squamous-cell carcinomaand adenoid cystic carcinoma as well as othertypes of lesions. One patient died of recurrentdisease two and one-half years following resection.There was one postoperative death which wouldnow be avoidable. Postoperative irradiation hasbeen added in one patient with a high adenoidcystic carcinoma of the trachea who requiredbevelling of the lower part of the larynx. Althoughfrozen sections taken of the retained tissue at thetime of resection showed no residual tumour, themargins were small enough so that irradiationwas added. Many of these lesions are radio-sensitive.

TABLE VITRACHEAL STENOSIS: COMPLICATIONS OF RESECTION

(17 PATIENTS)

Suture line separation. 2Suture line granulations. 8Partial restenosis.5Wound infection. 3Sternal separation 1Innominate haemorrhage .. .. 1Obstruction: adjacent malacia .. 1

cricoidal defect 1Aspiration, persistent. 1Vocal cord dysfunction. .. 2

In this series, 17 patients have had complica-tions of varying severity (Table VI). Suture lineseparation occurred in two patients. In one itwas minor posterior partial separation whicheventually healed well. In the second it followedhaemorrhage from the innominate artery, prob-ably due to intraoperative injury three daysfollowing the original operation. This led to sepsisin the wound, exposure of the anastomosis, andtracheal leakage after one week. Ultimately thishealed with a partial stricture which was function-ally acceptable. Granulations have appeared atthe site of the suture line in eight patients, andmost of these have been managed by broncho-scopic removal of granulations and sutures. In fiveof these patients there was partial restricturingwhich was often difficult to separate from theformation of granulations. In each of thesepatients conservative management of the granu-lations with or without direct intra-anastomoticinjection of triamcinolone has led to resolutionand a satisfactory functional airway.

Specific pre-, intra-, and short-term postopera-tive antibiotics have been used dependent uponthe specific sensitivity studies of sputum andstomal cultures. Wound sepsis, despite the presenceof open stomas and bacteriologically resistant

677

Hermes C. Grillo

Staphylococcus aureus or Pseudomonas aeruginosahas been infrequent. In one young patient withnormal tracheal flora antibiotics were not usedand staphylococcal superficial wound infectionoccurred. In one patient with abdominal woundsepsis resected as an emergency the sternumbecame infected but not the trachea. In a thirdpatient sternal separation seemed to precede localinfection. Obstruction due to the failure to resectan adjacent malacia or to correct a cricoid defecthas been mentioned. One patient who had per-sistent neurological difficulty following aspirationhas been described. In two patients there wasevidence of unilateral vocal cord dysfunctionwhich may or may not have been present in onepatient before the operation.

CONCLUSION

This experience demonstrates that we now havemethods for the effective surgical treatment ofthe majority of tracheal lesions whether neoplasticor inflammatory. Lesions involving up to andmore than 50% of the trachea may be effectivelyremoved and in almost every case primary recon-struction may successfully be done. With thispotential for treatment it is hoped that primarytumours, although rare, will be recognized morepromptly and treated more adequately in the initialoperation. We have seen no patient where re-operation on a tracheal carcinoma has led to apermanent cure. The evolution of our knowledgeabout iatrogenic lesions which follow respiratorytherapy and the development of methods of pre-vention should reduce the incidence of theselesions. On the other hand, with the extension ofeffective respiratory care and increasing survivalrates, stenoses will remain a problem.

REFERENCESAndrews, M. J. and Pearson, F. G. (1971). The incidence

and pathogenesis of tracheal injury following cuffedtube tracheostomy with assisted ventilation: an analysisof a two-year prospective study. Annals of Surgery,173, 249.

Barclay, R. S., McSwan, N., and Welsh, T. M. (1957).Tracheal reconstruction without the use of grafts.Thorax, 12, 177.

Belsey, R. H. R. (1950). Resection and reconstruction of theintrathoracic trachea. British Journal of Surgery, 38,200.

Cantrell, J. R. and Folse, J. R. (1961). The repair of circum-ferential defects of the trachea by direct anastomosis:Experimental evaluation. Journal of Thoracic andCadiovascular Surgery, 42, 589.

Cooper, J. D. and Grillo, H. C. (1969a). The evolution oftracheal injury due to ventilatory assistance throughcuffed tubes. A pathologic study. Annals of Surgery,169,334.

4 (1969b). Experimental production and preventionof injury due to cuffed tracheal tubes. Surgery, Gyneco-logy and Obstetrics, 129, 1235.

Dedo, H. H. and Fishman, N. H. (1970). Laryngeal releaseand sleeve resection for tracheal stenosis. Annals ofOtology, Rhinology and Laryngology, 78, 285.

Florange, W., Muller, J., and Forster, E. (1965). Morphologiede la necrose tracheale apres tracheotomie etutilisation d'une prosthese respiratoire. Anesthesie,Analgesie, Reanimation, 22, 693.

Geffin, B., Bland, J., and Grillo, H. C. (1969). Anestheticmanagement of tracheal resection and reconstruction.Anesthesia and Analgesia . . . Current Researches, 48,884., Grillo, H. C., Cooper, J. D., and Pontoppidan, H.(1971). Stenosis following tracheostomy for respiratorycare. Journal of the American Medical Association, 216,1984.

Grillo, H. C. (1965). Circumferential resection and recon-struction of the mediastinal and cervical trachea.Annals of Surgery, 162, 374.

- (1966). Terminal or mural tracheostomy in the anteriormediastinum. Journal of Thoracic and CardiovascularSurgery, 51, 422.(1969a). The management of tracheal stenosis followingassisted respiration. Journal of Thoracic and Cardio-vascular Surgery, 57, 52.(1969b). Surgical approaches to the trachea. Surgery,

Gynecology and Obstetrics, 129, 347.(1970). Surgery of the trachea. In Current Problems in

Surgery. Year Book Medical Publishers, Chicago(monograph July issue).(1972a). Tracheal anatomy and surgical approaches.

In General Thoracic Surgery, edited by T. W. Shields,chapter 30. Lea and Febiger, Philadelphia.(1972b). Benign and malignant diseases of the trachea.

In General Thoracic Surgery, edited by T. W. Shields,chapter 31. Lea and Febiger, Philadelphia., Bendixen, H. H. and Gephart, T. (1963). Resection ofthe carina and lower trachea. Annals of Surgery, 158, 889., Cooper, J. D., Geffin, B., and Pontoppidan, H. (1971).A low-pressure cuff for tracheostomy tubes to minimizetracheal injury. Journal of Thoracic and CardiovascularSurgery, 62, 898., Dignan, E. F. and Miura, T. (1964). Extensive resec-tion and reconstruction of mediastinal trachea withoutprosthesis or graft: An anatomical study in man.Journal of Thoracic and Cardiovascular Surgery, 48, 741.,' ,~9 ~~ (1966). Experimental reconstruction ofcervical trachea after circumferential resection. Surgery,Gynecology and Obstetrics, 122, 733.

Hajdu, S. I., Huvos, A. G., Goodner, J. T., Foote, F. W.,Jr., and Beattie, E. J., Jr. (1970). Carcinoma of thetrachea. Clinico-pathologic study of 41 cases. Cancer,25, 1448.

Harley, H. R. S. (1971). Laryngotracheal obstruction com-plicating tracheostomy or endotracheal intubation withassisted respiration. Thorax, 26, 493.

Houston, H. E., Payne, W. S., Harrison, E. G., Jr., andOlsen, A. M. (1969). Primary cancers of the trachea.Archives of Surgery, 99, 132.

Macmillan, A. S., Jr., James, A. E., Jr., Stitik, F. P., andGrillo, H. C. (1971). Radiological evaluation of post-tracheostomy lesions. Thorax, 26, 696.

678

Reconstruction of the trachea

Miura. T. and Grillo, H. C. (1966). The contribution of theinferior thyroid artery to the blood supply of thehuman trachea. Surgery, Gynecology and Obstetics, 123,99.

Mulliken, J. B. and Grillo, H. C. (1968). The limits oftracheal resection with primary anastomosis. Furtheranatomical studies in man. Journal of Thoracic andCardiovascular Surgery, 55, 418.

679