Fifty years ago pulmonary tuberculosis was the entity that defined thoracic surgery as a surgical specialty. Today, the debates over the various forms of collapse therapy, collapse therapy as an alternative to bed rest in a sanatorium, resection versus collapse therapy, and if resection, how extensive, are over. This changed scenario in the treatment of tuberculosis is clearly attributable to the development of effective antibiotics and chemotherapy rather than to technical advances, adjuvant support systems, or increased understanding of the disease. There is probably less understanding of the objectives of surgical treatment of tuberculosis now than in the past, but in spite of enormous changes in the social, national, and international aspects of tuberculosis, it remains a disease to be reckoned with.

The objective of surgery in the treatment of pulmonary tuberculosis is to promote stable healing or to remove an unhealed focus of disease: either a residual pulmonary cavity or a thick walled empyema space. The term ‘cavity’ is indicative of destroyed lung tissue or lobe that is subject to repeated tuberculous or pyogenic infections. Empyema requires excision by pleuropneumonectomy rather than drainage. Pulmonary cavities persisting after a course of chemotherapy that has rendered the sputum negative for acid-fast organisms—so-called ‘open negatives’ are no longer considered indications for surgery.

There has been an interesting cycle in the indications for resection. Before effective chemotherapy was available, indications for resection were so strict that few were undertaken. When chemotherapy made cutting through contaminated, but viable, tissue safe, limited resections became popular. Long-term studies then demonstrated that chemotherapy was so effective that resection was unnecessary even for cavities, and resection is now limited to relative indications, ‘calculated risks,’ in the accepted jargon.

The relative indications for resection are, first, an open negative cavity more than 2 to 3 cm in diameter, in a good risk patient who may be subject to stress in the future. This indication may apply to the immunologically compromised host. Second, the demonstration of atypical, drug-resistant organisms in cavitary disease that can be resected cleanly by lobectomy. Third, recurrent, sputum-positive infection in a pulmonary segment or lobe, even though no macroscopic cavity is clearly demonstrated, if one accepts the classic dogma that a positive sputum is a manifestation of cavitary disease, and, fourth, to exclude malignancy, absit fine needle biopsy, in an asymptomatic peripheral nodule of uncertain duration. If the nodule is tuberculosis, a so-called ‘tuberculoma’, limited resection is usually curative even without postoperative chemotherapy, and elective removal may therefore be advisable.

The inexact term ‘tuberculoma’ refers to cavity filled with inspissated, caseous material distal to a blocked bronchocavitary junction. This is unstable cavity healing for, theoretically, the bronchial blockade may dissolve, the caseous material may liquefy, and endogenous reinfection may occur.

The term ‘cavity’ is so prominent in any discussion of the surgical treatment of tuberculosis, that some concept of its natural history is important. The names Pagel, Coryllos, Alexander, Pinner, Klopstock, Churchill and many others flicker on the historical horizon. Few diseases comapre with tuberculosis in the matter of subjective interpretation, from pathologists to poets. What follows is my personal synthesis derived from reading when I took the ‘cure’, including artificial pneumothorax, far away and long ago before chemotherapy was available.

Stage I of pulmonary tuberculosis is the result of initial infection by air-borne organisms carried in fine droplets or dust into the peripheral bronchial tree. The size of the infecting dose and the virulence of the organisms influence the intensity and extent of the host reaction. This may vary from a trivial, undetected pneumonitis to the lethal pneumonia seen in infants, malnourished children in war-torn or impoverished environments, native populations traditionally free of tuberculosis, and immunosuppressed individuals. In this stage (primary or childhood tuberculosis) the patient is tuberculin negative and the disease does not progress to an intense, necrotizing process, or abscess cavity. The majority of patients, however, become positive to the tuberculin trest and may develop an insignificant scar or Ghon complex. In some patients, Stage I tuberculosis may progress to chronic infection of the lymph and haematogenous systems (Stage II). In this construct, Stage I is not cavitary disease, and is therefore without surgical implications, but it should be treated by chemotherapy.

Stage II tuberculosis is the lymph-haematogenous or disseminated form of the disease, and its most striking and lethal, but fortunately uncommon, manifestation, is disseminated miliary tuberculosis. As a rule there are specific targets of haematogenous spread: tendon sheaths, bone and joints, kidneys, meninges, pericardium, lung (effusion and tuberculous empyema), lymph nodes, and soft tissues (cold abscesses and Pott's disease), as well as pulmonary infiltrates that do not become cavities except as a terminal phenomenon. One expects to see stage II tuberculosis in patients with AIDS. This stage is also ‘non-surgical,’ except when surgery is appropriate for non-pulmonary targets. Certain patients never progress beyond Stage II, yet do well clinically for years with chemotherapy and an improved environment.

Stage III is, however, an object of our surgical concern, and is represented by cavitary disease, phthisis, ‘consumption’, or ‘adult tuberculosis’. In this stage, the patient has acquired a marked sensitivity to the tubercle bacillus so that reinfection of the lung, whether from an exogenous or endogenous source, results in inflammatory fixation and intense tissue destruction—an abscess surrounded by inflammatory infiltrate. Reinfection may result from a new exogenous contact, or endogenous spread from a preexisting cavity or erosion of a lymph node into a bronchus. Physical, nutritional, psychological and immunological competence of the host, the invasive and destructive capability of the infecting strain, and other factors are important modifiers of this stage of the disease process.

The initial treatment goal in adult tuberculosis Stage III caused by reinfection is control of the invasive component. Before chemotherapy this meant bed rest, sanatorium care, social service support, fresh air, sunshine, and hope. The dramatic change wrought by chemotherapy was to reduce the time taken for cure (particularly important for young people) and environmental demands to the point of substituting ambulatory care for sanatoria. It also reduced the number of frustrating failures of medical cure; unhealed, positive cavities which remained after the surrounding infiltrate had disappeared.

Since cavities represent destruction of tissue, stable healing means the formation of a dense, linear, avascular scar in which organisms either cannot survive or are densely encapsulated. Such scars are familiar findings in routine autopsies and in patients who have been cured or who are chronically infected. The sequential moves of collapse therapy: phrenicectomy—artificial pneumothorax—internal pneumonolysis—extrapleural pneumothorax and plombage—thoracoplasty, aimed to hasten production of such scars. Relaxation of the elastic tissue in the lung, which prevented scar contracture, was the rationale for collapse therapy. Collapse therapy is now obsolete except as an adjunct to drainage of chronic empyema cavities when decortication is not an option.

Unstable healing of cavities may result from blockage of the cavitary–bronchial junction followed by the accumulation of inspissated caseous material, resulting in a ‘tuberculoma’. When multiple, the process is called nodular, or fibronodular tuberculosis.

A third method of cavity healing, now considered relatively stable through without closure, is extension of healthy epithelium from the draining bronchus to line a residual cavity—, ‘an open negative.’ Prior to chemotherapy it had been repeatedly demonstrated that viable tubercle bacilli might persist for years in the walls of such cavities. Now chemotherapy effectively controls the organisms until they are enclosed by dense scar tissue.

In the unusual situation where resection is thought to be indicated, certain technical points are important. Resection should be covered by chemotherapy beyond the usual first-line agents, isoniazid, ethambutal, rifampin, and pyrazinamide if the offending organism has become resistant. Secondly, bronchoscopy should disclose little or no evidence of active inflammation that might compromise healing at the level of the proposed bronchial resection. Thirdly, a maximum amount of uninfected lung tissue should be preserved; hence, segmental resection though the hilar structures are anatomically distorted by dense scar. Finally, the axiom that primary healing is the objective of surgery for tuberculosis translates into meticulous use of the electrocautery or fine ligatures and the pleuralization of raw surfaces to avoid continuing air leaks. Resection for pulmonary tuberculosis should be highly selective; technical misadvantures resulting in extensive excision may be disastrous.

Surgery has an important place in the management of tuberculous empyema, an uncommon but important manifestation of the disease which should be distinguished from non-tuberculous pleural space infection.

A minority of patients with tuberculous empyema have no evident bronchopleural fistula and can be managed by evacuation, decortication, and expansion of an underlying lung which has not been badly damaged by previous disease.

The majority of patients with tuberculous empyema, however, have an associated bronchopleural fistula with secondary infection and an irrevocably damaged lung. The individual is likely to be chronically ill, malnourished and depleted. In these patients the first step is to be certain there is adequate, dependent drainage of the empyema to minimize further aspiration of its contents into the bronchial tree with insult to the contralateral side. Secondly, attention should be directed to the nutritional and physical status of the patient.

When a plateau of well-being is reached, pleuropneumonectomy or pleurobectomy is the procedure of choice. Sarot defined the rationale and technique of this operation in 1949, just as bacteriological control became a reality, though the concept of excision and primary closure had been long established in surgery for tuberculosis.

The procedure requires a generous posterolateral thoracotomy including the previous drainage site, and development of the extrapleural space, allowing the operator to excise as completely as possible the empyema cavity from the chest wall and diaphragm. Dissection is then carried into the mediastinum to isolate and to divide the hilar structures. The hilar dissection is not usually difficult, for the major pulmonary vessels are reduced in size and the peribronchial tissues, unless there has been an associated mediastinitis, are relatively avascular. Closure of the bronchus must be performed carefully with as short a stump as possible. The suture line must be reinforced with viable mediastinal tissue or an intercostal pedicle flap. The chest is then closed without drainage or, at the most, intercostal suction drainage for 8 to 12 h. Appropriate pre-, intra-, and postoperative chemotherapy is essential; local instillation of antibiotics is optional.

This formidable operative procedure is well tolerated when there is reasonable reserve on the contralateral side and the patient is in reasonable balance with his disease; long-term results are equally good. The Achilles heel is primary healing of the bronchial stump.

For centuries the treatment of infected fluid in the pleural space has been the business of surgery, but only the last century or so has the approach been rational and physiological. The definition of what is now called the classic sequence in the treatment of empyema, that is, aspiration—closed drainage—open drainage, is attributed, in the United States at least, to Evarts Graham and the Empyema Commission of 1918, appointed during the influenza pandemic and the last year of the First World War. In actual fact, the same principles had been well described in the German and english surgical literature. The Second World War established the value of an alternate approach—decortication, and from the late 1940s, antibiotics and chemotherapy brought about a synthesis of all the above.

In the treatment of empyema thoracis, the analogy of a tree surgeon faced with a rigid walled cavity is useful. Short of cutting down the tree, he solves the problem of eliminating dead space by filling the cavity with cement after removing any decay. Since we do not have the option of leaving a large foreign body in the presence of infection, our cement must be autogenous, viable tissue, such as expanded lung, muscle and skin flaps, omental grafts, or the soft tissues of the chest wall. From a functional point of view, the preferred alternative is an expanded, reasonably normal lung. Both the classic sequence and decortication serve that purpose, with antibiotics allowing modifications of both methods and shortening the time frame in either case.

Non-tuberculous empyema, which is our concern in the following paragraphs, may be a complication of acute reversible infection in the underlying lung (classically, but now only rarely, lobar pneumonia of pneumococcal or streotococcal origin), chronic bronchopulmonary suppuration, bronchiectasis or lung abscess, or infection of a traumatic haemothorax, contaminated either at the time of injury or later, and associated with varying degrees of reversible pulmonary damage.

The pathogenesis of synpneumonic and postpneumonic empyema has been well defined. An initial pleural effusion of low fibrin content and low specific gravity may, in the absence of antibiotics, or in spite of antibiotics, increase not only in volume, but also in fibrin content. Unless removed by multiple aspirations, (or by closed suction tube drainage when the fluid becomes too viscous to aspirate) fibrin condenses on the visceral pleura to entrap the lung within a restricting envelope. Were rib resection and open drainage to be carried out at this stage, the lung would collapse, as with any open pneumothorax. This was the disastrous scenario when early drainage was advocated for the streptococcal effusions prevalent in the influenza pandemic of 1918.

Prior to the development of antibiotics, it was unusual for multiple aspirations to be adequate to keep the lung totally expanded long enough for the visceral and parietal pleural surfaces to become adherent, or for closed, tube section drainage to succeed in eliminating any residual purulent collections. Indeed, a manageable and accessible residual space, in a dependent region close to the costophrenic angle, with most of the lung expanded and adherent firmly to the chest wall, was the objective of closed drainage. When adherence was judged to be stable—a matter of a week or 10 days—resection of a short segment of rib and open drainage were indicated. This classic sequence is still useful when antibiotics fail to interrupt its course, or in elderly or debilitated patients who cannot tolerate thoracotomy and decortication. With open drainage, final healing with obliteration of the cavity may be a slow and sometimes painful process, weeks and often months of devoted wound care being required to assure that the drainage tube remains in place and that drainage is dependent and continuous.

The alternative to open drainage is a mjor thoracotomy, removal of the fibrinous, visceral, pleural peel and re-expansion of the lung under direct vision (decortication) followed by a short period of closed suction drainage until air leaks seal and the lung remains expanded. Successful decortication depends on selecting antibiotics against the offending organism so that invasive infection is controlled, and on the presence of sufficient expandable underlying lung, which in turn depends upon the primary disease of which the empyema is a complication.

Lobar pneumonia is usually rapidly reversible by antibiotics; underlying bronchiectasis and lung abscess, on the other hand, are not. In the latter the surgeon must consider concomitant resection of the primary process; an analogy is the resection of a diseased appendix along with the removal or evacuation of localized peritonitis. In the case of an infected haemothorax, the extent and nature of the trauma to the underlying lung must be carefully assessed. The integrity of the bronchial tree is of critical importance, since at the conclusion of the decortication, expansion of the lung will be achieved by positive intratracheal pressure. Preliminary bronchoscopy is essential.

The management of empyema complicating carcinoma of the lung may present a difficult dilemma. Categorically, cancer of the lung that has gone on to pleural infection is incurable, though anecdotal exceptions occur, but incomplete resection may be palliative. The patient needs to know that morbidity in these circumstances is predictably high. There are occasions in which external drainage of profuse, malodorous discharge is preferable to a distressing, productive cough—a difficult judgement to make.

Chronic, non-tuberculous empyema is often associated with a persisting bronchopleural fistula, which must be dealt with in addition to, or as part of, space obliteration. One approach involves wide opening of the empyema space (appropriate antibiotic coverage is assumed), identification of the fistula, circumcision of the surrounding dense scar, and closure of the underlying bronchus in flexible mediastinal or pulmonary tissue. The closure is then reinforced by a pleural or intercostal bundle flap, or by thoracoplasty if insufficient lung is available as in a post-pneumonectomy patient.

Subacute or chronic empyema in the postpneumonectomy patient in the absence of a bronchopleural fistula, merits serious consideration of a Clagett sequence of procedures. The first stage is open rib resection and drainage with construction of an Eloesser flap—an inverted U-shaped skin flap lining of the drainage tract that eliminates the need for a thoracostomy tube. Topical irrigations of the empyema space are then performed for 2 or 3 weeks until the fluid is relatively clear. The patient is then returned to the operating room, the cavity filled with a dilute solution of neomycin (Stafford and Clagett suggested 0.25 per cent in normal saline), and the drainage tract is closed in layers. The object is, of course, to avoid extensive thoracoplasty in patients who have had a pneumonectomy for cancer, with all the uncertainty that that implies. Modifications of the sequence come with experience.

The key to management of chronic empyema is closure of the bronchial communication. Under special circumstances, such as a post-right pneumonectomy bronchopleural fistula with an excessively long right main bronchial stump, one may elect a trans-sternal approach as a first stage, followed by thoracoplasty, space filling muscle flaps, and Clagett procedures, to fill the space, which is the surgical equivalent of the tree surgeon's cement of our earlier metaphor.

Alexander J. The collapse therapy of pulmonary tuberculosis. Springfield, IL: CC Thomas, 1937.

Churchill ED, Klopstock R. Lobectomy for pulmonary tuberculosis. Ann Surg 1943; 117: 641.

Eloesser L. An operation for tuberculous empyema. Surg Gynecol Obstet, 1935; 60: 1935.

Ginsburg RJ. Median sternotomy for repair of B-P fistula postneumonectomy. In: Martini N, Vogt-Moykopf I, eds. International trends in general thoracic surgery. St. Louis: CV Mosby, 1989: 149.

Graham EA, Bell RD. Open pneumothorax; its relation to the treatment of acute empyema. Am J Med Sci 1918; 156: 83.

Sarot IA. Extrapleural pneumonectomy and pleurectomy in pulmonary tuberculosis. Thorax 1949; 4: 173.

Stafford EG, Clagett OT. Postpneumonectomy empyema. J Thoracic Cardiovasc Surg, 1972; 63: 771.