Primary chest wall tumours account for approximately 1 to 2 per cent of all neoplasms and approximately 7 to 8 per cent of all bony neoplasms. The most common primary chest wall malignancies are those of bony or cartilaginous origin, followed by soft tissue sarcomas. Most bony lesions occur in the ribs; sternal tumours are rare and generally malignant. Approximately 50 per cent of chest wall tumours are malignant, however, their incidence varies from 9 to 80 per cent, depending on the series (Table 1) 534. In a series of patients with 48 primary chest wall tumours there were 30 rib lesions and 4 sternal tumours; the remainder originated in the soft tissues of the chest wall. Twenty-six of the 48 tumours were malignant (54 per cent); all sternal tumours were malignant.

Because primary chest wall tumours are uncommon, few large series have been published. It is difficult to determine the proportion of chest wall masses that are primary tumours rather than metastatic lesions; however in a series of 100 consecutive chest wall resections, 60 per cent were performed for primary chest wall tumours; the others were for metastatic or locally invasive tumours.

The most common benign primary lesions in most series are fibrous dysplasia and chondroma; chondrosarcoma is the most common primary chest wall malignancy. Although there are wide differences in reporting, other common malignant tumours include malignant fibrous histiocytoma, rhabdomyosarcoma, fibrosarcoma, and Ewing's sarcoma (Table 2) 535.

Chest wall tumours frequently present as enlarging palpable masses or as pain, although signs and symptoms are absent in 10 to 20 per cent of patients. Painful tumours tend to be malignant. Asymptomatic lesions are often detected on radiographs obtained for routine screening or following trauma.

While history and physical examination may be helpful in establishing a diagnosis, the definitive diagnosis is generally made after further radiographic investigation and histological analysis. Patients with chest wall masses should undergo routine laboratory screening tests, including alkaline phosphatase, serum calcium, tests of renal function (creatinine and urea or BUN), and, when myeloma is suspected, serum protein electrophoresis, to detect evidence of systemic or metastatic disease.

The differential diagnosis of chest wall masses includes non-specific chondritis (Tietze's syndrome), costochondral separation, congenital or developmental deformities such as pectus carinatum, and tuberculous chondritis, which is becoming less common.

Accurate radiographic assessment is essential in the evaluation of chest wall masses, especially those of bony or cartilaginous origin. Some lesions, such as chondrosarcoma, have distinctive radiographic characteristics, but many lesions yield non-specific radiographic findings. Standard posteroanterior and lateral chest radiographs should be obtained and supplemented with rib views. CT is useful in determining the origin and extent of the lesion, involvement of adjacent structures, and presence of pleural effusions, but it has not been shown to be of great benefit in establishing an initial diagnosis. CT scans are the most sensitive method of detecting pulmonary metastases.

Radionuclide bone scans are useful in detecting multiple lesions, spread to adjacent ribs, and when multiple myeloma is suspected. Angiography can be useful in planning myocutaneous flaps for reconstruction of large chest wall defects and may be useful in assessing tumours near the axillary vessels. The role of magnetic resonance imaging in evaluating chest wall masses has not been defined.

While clinical evaluation and radiographic investigations are important in managing chest wall neoplasms, accurate histological analysis is the single most important factor in determining the appropriate treatment. There is controversy regarding the most appropriate method of obtaining tissue for histological analysis: needle biopsy, incisional biopsy, or excisional biopsy.

Needle biopsy, while the least invasive, is the least informative and can produce misleading results. It is useful in the evaluation of lesions suspected of being either metastatic tumour deposits or multiple myeloma.

Incisional biopsy generally provides the correct diagnosis but is not curative. Tumour seeding along the biopsy tract can occur, and misleading information may be obtained from large heterogeneous lesions. Nonetheless, this technique is very useful when dealing with large lesions or tumours that are best treated with preoperative chemotherapy or radiotherapy. An incisional biopsy must be carefully planned so that it can be encompassed by a future, more extensive resection. When performing an incisional biopsy for a small lesion, tumour localization may be difficult; the use of bone scans to identify and localize the lesion for placement of a skin marker can be helpful. The rib can also be injected with methylene blue after needle localization. When these techniques are used the patient should be biopsied in the same position in which the scan was obtained.

Excisional biopsy has the obvious advantage that it is curative for most benign lesions and for some small malignancies. Our practice has been to perform excisional biopsy on small lesions and incisional biopsy on larger tumours. If the tumour is malignant or incompletely resected, definitive resection is undertaken as a second procedure. This two-stage approach avoids the use of unnecessarily wide excisions for certain tumours and permits coordination between multiple specialties, such as radiotherapy and plastic surgery, before an extensive resection is undertaken.

Biopsy samples should be sent for routine histopathological analysis, although more specialized immunohistochemical techniques or electron microscopy are sometimes helpful. Frozen sections are rarely used for making an initial diagnosis although they may be helpful in determining whether the biopsy specimen contains an adequate amount of viable specimen for evaluation.

The histology of lesions from the recent experience of the Massachusetts General Hospital, which is similar to most reported series, is shown in Table 3 536. Most variations from other reports can be attributed to the rarity of these tumours, resulting in relatively small numbers of patients, the inhomogeneity of tumour types, the relatively long time frames over which data were collected, geographic locations from which patients were selected, and patterns of referral to the reporting institution.

Fibrous dysplasia is a common lesion that often presents as chest wall pain. It is either monostotic or polyostotic as a component of Albright's syndrome, and the rib cage is the most frequent site of this slowly growing lesion. While Adkins claims that fibrous dysplasia generally presents in the posterior or lateral portions of the rib, we have found the lesion to be equally distributed around the rib. Radiographically, the lesion demonstrates cortical expansion and thinning with a trabeculated lytic centre. In two of eight patients at this institution the radiographic diagnosis, made prior to excision, was at variance with the pathological diagnosis; the incorrect diagnoses were enchondroma and aneurysmal bone cyst. Since the results of needle biopsy are also unreliable, the diagnosis should be made by excisional biopsy. Microscopically, the lesion consists of fibrous tissue with poorly oriented and incompletely mineralized bone trabecula. Resection of the lesion provides symptomatic relief and recurrence is rare.

Eosinophilic granuloma is not a true benign neoplasm, but behaves like one in its presentation, clinical course and treatment. Eosinophilic granuloma of bone is not generally associated with diffuse visceral disease, unlike histiocytosis X, Letterer-Siwe disease, or Hand-Schuller-Christian disease, nor are there generally any systemic symptoms, although the lesion usually causes some pain. Radiographically this lesion appears as an expansile, lytic lesion with cortical thinning and periosteal new bone formation. The microscopic appearance is one of a reticulum of histiocytes infiltrated with eosinophilic leucocytes which may have bilobed or hyperchromatic neclei. The eosinophilic cells may be missing and replaced by lymphocytes, plasma cells, or neutrophils. Eosinophilic granuloma is cured by resection of the affected rib, curettage, radiation, or by one of several antimetabolic drugs.

This rare benign tumour usually arises adjacent to the inferior angle of the scapula. The lesion, which is frequently asymptomatic, is generally considered to be a disease of the elderly, but has been reported in a patient as young as 6 years. The chest radiograph shows a soft tissue mass without rib involvement; CT often demonstrates an unencapsulated subscapular mass. Clinically, the lesion may resemble a malignant tumour being densely adherent to the muscle, rib periosteum, or scapula. On gross inspection, the lesion is a fibrofatty mass without a defined capsule; the microscopic appearance is of hypercellular fibrous tissue with irregular, thick, wavy elastic fibres that may require staining for elastic tissue to establish their identity. Since the treatment of elastofibroma dorsi consists of simple excision it must be differentiated from the more cellular and pleiomorphic low-grade fibrosarcoma or desmoid tumour, which requires a much wider resection margin.

Desmoid tumours are frequently located in the shoulder or chest wall and are benign lesions that originate in the fascia or muscle. They may invade local tissue and are felt to be either a form of fibromatosis or a low-grade fibrosarcoma. These lesions have a high incidence of local recurrence and should be widely excised or treated with enucleation and radiation.

Chondromas (or enchondromas) are hamartomas of cartilage. They account for 15 to 43 per cent of benign rib tumours, generally occur in young adults, and are frequently located at the costochondral junction. Radiographically, they are well defined, ovoid, expansile lesions with diffuse or stippled calcification. The histological appearance is one of cartilage with variable cellularity. The differentiation between chondroma and chondrosarcoma on clinical or radiographic findings can be impossible; all suspected chondromas should be treated by excision, as though they were a chondrosarcoma.

Chondrosarcoma is the most common primary malignant tumour of the chest wall, accounting for 20 to 50 per cent of all primary chest wall malignancies. Most chondrosarcomas arise de novo. Although malignant degeneration of previously benign lesions may occur, chondrosarcomas are unusual in patients under 30 years of age. Ipsilateral major chest wall trauma has been reported in 12.5 per cent of patients with rib chondrosarcoma. The importance of trauma in the aetiology of this condition is not clear, however.

A recent extensive review of chest wall chondrosarcomas included 55 men and 41 women. The mean age was 53.5 years (range, 17–78). Most of the tumours (81 per cent) arose in the ribs; 18 per cent originated in the sternum. Physical signs or symptoms were present in 94 per cent of patients, with more than 50 per cent having had signs or symptoms for longer than 1 year. Twelve patients had a history of major ipsilateral chest wall trauma.

The diagnosis of chondrosarcoma is best made by pathological evaluation after excision since the clinical signs, radiographic characteristics, and histological appearance of incisional biopsies may be identical to those of benign cartilaginous tumours. The radiographic appearance of chondrosarcomas arising in flat bones such as ribs or the sternum is usually one of a soft tissue mass with bony destruction and varying amounts of new bone formation. There is often mottled calcification and the soft tissue component may be large in proportion to the size of the skeletal lesion. Mesenchymal chondrosarcoma may not be associated with calcification due to the rapid growth rate of the tumour. The tumour is generally a lobulated firm mass that appears cartilaginous; its pathological and histological characteristics have been extensively described elsewhere. Microscopically, chondrocytes may be multinucleate or undergoing mitosis; they may be atypical, with large, heavily chromatized nuclei. Areas of myxomatous change may be identified. While difficult to grade histologically, chondrosarcomas are graded from I to III based on their malignant appearance as defined by pleiomorphism, cellularity, and nuclear atypia.

The natural history of most chondrosarcomas is one of slow growth and local recurrence, with the exception of the relatively uncommon high-grade lesions which grow quickly and metastasize early. The long-term prognosis is related to tumour grade, size, and extent of resection (Table 4) 537. Of the 72 patients in one series, 66 had borderline, Grade I, or Grade II lesions. The overall 10-year actuarial survival rate was 65 per cent. Local recurrence developed in 51 per cent of patients treated by resection.

The treatment of chondrosarcoma is based on surgical resection; few data have demonstrated benefits from either adjunctive radiotherapy or chemotherapy, but their use in patients with high-grade tumours may be warranted as the prognosis following resection of these lesions is so poor.

This unusual tumour, seen primarily in young patients, accounts for 8 to 24 per cent of all malignant chest wall tumours. Of all cases of Ewing's sarcoma in the combined experiences of the National Cancer Institute, the MD Anderson Hospital and the Mayo Clinic, 6.5 per cent were primary rib lesions. In one review of 36 patients with primary Ewing's sarcoma of the ribs, 21 had localized disease at the time of diagnosis, eight had regional disease, and seven had metastatic disease. The patients ranged from 4 to 25 years of age (mean, 13); there were 12 men and nine women. The predominant symptom was chest pain and symptoms had been present for 5 to 235 days (mean 36) days prior to evaluation. Although the tumour is highly radiosensitive, long-term survival following either surgery or radiation alone is poor. The addition of chemotherapy, consisting of cyclophosphamide, vincristine, doxorubicin (Adriamycin) and dactinomycin to radiotherapy and surgery improved survival, with a 3-year local control rate of 95 per cent and disease-free survival rate of 50 per cent. Trials of total body irradiation and bone marrow transplantation have also been undertaken.

This is generally considered to be a rare primary chest wall tumour, although it accounted for 17 per cent of all primary chest wall malignancies and 40 per cent of soft tissue sarcomas in the experience of the Massachusetts General Hospital. The tumour is composed of malignant fibroblasts arising from the medullary cavity (central fibrsarcomas) or periostium (cortical fibrosarcoma). The radiographic characteristics are that of a radiolucent lesion with little periosteal reaction. The margins of low-grade lesions are generally well defined, but those of high-grade tumours are diffuse. The diagnosis of fibrosarcoma is based on microscopic analysis of biopsy specimens, which show interlacing bundles of collagen without bone or osteoid formation. The lesions are graded from I to III based on their malignant appearance. The treatment of fibrosarcomas is similar to that of most soft tissue sarcomas, generally consisting of wide local excision for low-grade lesions, supplemented with radiation for high-grade tumours.

In our series of 48 primary chest wall malignancies there were eight fibrosarcomas. The average age of patients with this lesion was 50 years (27–85); there were three anterior or lateral soft tissue chest wall lesions, two presternal lesions, one sternal lesion and two in the upper back (included because they extended down to the ribs). Signs or symptoms, including the presence of a mass, had been present between 4 weeks and more than 15 years prior to presentation. Some tumours were quite large (20 × 10 × 5 cm). All patients were treated with wide excision; radiation was administered shortly after excision in two patients with spindle cell sarcomas (one with a positive resection margin) and three other patients after the development of local recurrence. Six of eight patients developed recurrent disease from 3 to 42 months after initial treatment. Two patients were alive and free of recurrence at 15 months and 7 years after initial treatment. The patient who was free of disease at 7 years had been treated only by local excision of a low grade lesion; the only other disease-free patient had undergone wide excision of a small (4 cm) high-grade (III/III) spindle cell sarcoma followed by 5000 cGy of external beam radiation and 1400 cGy by radium implant. Mean survival was 36 months; death occurred 7 to 41 months after initial treatment.

Rhabdomyosarcoma is a rapidly growing tumour of skeletal muscle that is generally seen in children and young adults. It is one of the more common soft tissue sarcomas and frequently occurs in the chest wall. This tumour is unique in that radiation and chemotherapy produce very good long-term survival—70 per cent at 5 years, although prior to the use of combination therapy long-term survival was poor. The role of surgery in this condition is controversial: Pairolero recommends wide excision while Pass maintains that surgery should be limited to generous biopsy and removal of residual disease after courses of radiation therapy and chemotherapy.

Other soft tissue sarcomas, such as liposarcoma, malignant fibrous histiocytoma, synovial cell sarcoma, leiomyosarcoma, rhabdomyosarcoma, malignant Schwannoma, and haemangiosarcoma may be found in the chest wall. As with fibrosarcoma, they are graded from i to iii based on their histological appearance, and this corresponds with prognosis. These tumours tend to recur locally and to metastasize to the lungs. With the exception of rhabdomyosarcoma, they should be treated by radical excision. The role of adjuvant chemotherapy and radiotherapy for truncal sarcomas has not yet been well defined because of the small numbers of patients with these lesions.

Solitary plasmacytomas may occur in the ribs; however most of these eventually prove to be multiple myeloma. Osteosarcoma of the chest wall is generally considered to be rare, although Adkins reported it to account for 15 per cent of malignant tumours of the bony thorax. Wide local excision is indicated, but there is evidence that adjunctive chemotherapy may be beneficial. Malignant neuroepithelioma resembles Ewing's sarcoma, but can be differentiated from it by staining for neurone-specific enolase or by demonstrating dense core granules by electron microscopy.

Most patients with chest wall neoplasms require surgical resection as part of their treatment. The extent of resection depends on the size and type of the neoplasm: benign lesions generally require only limited resection with clear margins. Malignant lesions usually require resection with wide margins. Opinions differ with respect to the extent of resection for malignant lesions. Pass has recommended complete resection of any involved rib with its cartilaginous articulations as tumour extension within the bone marrow cannot be predicted, while Adkins has advocated routine resection of the ribs above and below the involved ribs in order to gain adequate margins. Others recommend resection of the lesion alone, with a 4- to 5-cm margin of normal tissue; Pairolero has amended this to include removal of the entire involved rib for high-grade malignancies.

Analysis of frozen sections may be helpful in assessing resection margins, but may be difficult to interpret. For sternal or manubrial tumours, the entire bone and adjacent costal arches must be removed. Any adherent tissue, such as anterior mediastinal fat, thymus, pleura, or lung, should be removed en bloc with the chest wall specimen. The limits of resection should be determined by adequacy of margins rather than by attempts to preserve chest wall as extensive chest wall resections can be performed safely and with good functional and cosmetic results.

The incision chosen for resection is determined by the size and location of the tumour, and may be a standard thoracotomy or variant thereof. Vertical incisions for incisional biopsies, should be avoided because subsequent wide resection, which must include the biopsy incision, will be difficult. The exception to this general rule is that median sternotomy is appropriate for resection of sternal lesions or when resection of bilateral pulmonary metastases is planned. When planning the incision the anticipated type of wound closure, either by primary closure or chest wall reconstruction, should be taken into consideration.

Reconstruction of a chest wall defect can be difficult. The primary consideration is usually the size and location of the bony and soft tissue defect, although factors such as preoperative pulmonary function, previous chest wall surgery or radiotherapy, and cosmetic considerations can be important. Reconstruction of segments of the chest wall is performed to protect underlying structures, to obtain chest wall rigidity and fixation for effective respiratory effort, and to prevent flail segments, reduce paradox, and prevent herniation.

Small defects (those involving fewer than three ribs) can usually be closed primarily; those greater than 5 cm often require chest wall reconstruction as a large flail segment would otherwise result. Location is important. Posterior defects up to 10 cm in diameter can be tolerated as long as the scapula remains to provide stability. Generally, resection of ribs 1 to 4 posteriorly is well tolerated, but if resection also involves the fifth and sixth ribs, additional support is necessary to prevent the scapula from becoming caught below the lower ribs. Resection which includes the costal margin can be reconstructed by suturing the diaphragm to the lowest remaining rib; this is not possible if resection extends above the fifth rib. Either the diaphragm will not reach high enough or the remaining hemithorax will be too small. Defects involving the sternum should be reconstructed to protect underlying structures and to prevent the paradoxical motion that occurs with removal of the anterior costal attachments.

Preoperative assessment of patients requiring major chest wall resection should include pulmonary function testing since a pulmonary resection may be necessary and a period of mechanical ventilation may be required postoperatively. Preoperative treatment with Adriamycin may cause impaired cardiac function; bleomycin may cause pulmonary fibrosis. Patients with severe respiratory limitations may benefit from chest wall reconstruction following the creation of defects in any area, as the margin for error in these patients may be quite small and any further decrease in pulmonary reserve may be severely detrimental.

Patients who have undergone previous chest wall surgery or radiotherapy present special concerns in that the vascular supply to the tissue remaining after chest wall resection must be considered prior to chest wall resection. Good vascular supply is particularly important if prosthetic material is to be implanted; adjacent poorly vascularized or irradiated tissue may significantly increase the risks of infection.

Chest wall reconstruction is aimed at recreating the bony rigidity and soft tissue coverage of the native chest wall. Synthetic material and autogenous tissue, such as bone grafts and soft tissue flaps, have been used to meet these goals. Synthetic materials are used to provide both rigidity and protection of underlying structures. Historically, materials such as steel, tantalum, Ivalon, Lucite, and fibreglass have been used. Most of these create such a tissue reaction that they are rendered unusable; they have, therefore, been replaced by less reactive materials. Marlex mesh, which has been used since 1960, has been the standard synthetic material for reconstructing large defects. It can be used without additional support or can be reinforced with methyl methacrylate. This may be fashioned outside the chest wall defect or may be assembled in situ. The rigid methyl methacrylate component must be made smaller than the chest wall defect to provide a cuff of Marlex which can be sewn to the adjacent ribs. Too large a prosthesis may result in unnecessary pain and stiffness. Methyl methacrylate–Marlex mesh prostheses are well suited to reconstruction of sternal defects because they provide the necessary rigidity to protect the underlying heart and great vessels. Problems with Marlex and methyl methacrylate include a transient metabolic acidosis during the curing phase of the plastic, and seroma formation, although this has not been our experience.

More recently, 2 mm thick polytetrafluorothethylene (PTFE or Gore-Tex&supT;&supM;), polypropylene (Prolene&supT;&supM;) mesh, and Dexon&supT;&supM; mesh have been used with good results; some surgeons prefer them to Marlex. PTFE is impermeable to both air and water, is easy to work with, but lacks rigidity. Prolene differs from Marlex in that it has multidirectional rigidity; Marlex is rigid in only one direction.

Although infection is a potential problem when any synthetic material is used, these patches rarely need to be removed for the treatment of infection.

Prosthetic materials provide a rigid base for reconstructing chest wall defects, but the overlying soft tissue must also be replaced. The pectoralis major, latissimus dorsi, and rectus abdominus myocutaneous flaps, and greater omentum are all available for use.

The pectoralis major myocutaneous flap is based on the pectoral branch of the thoracoacromial artery. The pectoralis, taken as a muscle flap, or with its overlying skin as a myocutaneous flap, can be mobilized to provide coverage to the neck and upper sternum; it has frequently been used to provide closure for upper chest and neck wounds. It is also suitable for covering smaller defects of the anterior and lateral chest wall. The lateral tendinous attachments may be divided if additional mobility is necessary; it is also possible to mobilize the pectoralis muscle based on perforators from the internal mammary artery and to rotate the muscle medially to cover sternal defects. The pectoralis flap usually requires prosthetic material for underlying structural support and a split thickness skin graft to close the donor site.

The latissimus dorsi flap is based on the thoracodorsal artery, a branch of the subscapular artery. The flap, which is generally used as a myocutaneous flap, can support an island of skin up to 10 × 16 cm in area. It can cover areas larger than the pectoralis flap and, with a long vascular pedicle, can be mobilized to cover chest wall defects over a wide area. It may not, however, be able to reach to the inferior costal margin in the area of the xiphoid. A reverse latissimus flap has been described, based on branches of intercostal and lumbar arteries, in which the muscle is rotated medially to cover contralateral posterior defects. As with the pectoralis flap, skin grafting is usually required for closure of the donor site.

The transverse rectus abdominis flap, which is frequently used for breast reconstruction, is also excellent for covering large chest wall defects. This bipedicle flap, based on the inferior epigastric artery and superior epigastric artery, a branch of the internal mammary artery, can be mobilized on either of these vessels to cover defects as large as an entire anterior hemithorax from the costal margin to the clavicle. When used to cover defects created by removal of the sternum, the flap offers such stability that additional prosthetic material is not necessary. Three other muscle transposition flaps include the serratus anterior, trapezius, and external oblique flaps, all of which offer coverage to only limited areas of the chest wall.

Greater omentum, based on either the right or left gastroepiploic arteries, can be brought from the abdomen to almost any location in the thorax and used to provide soft tissue coverage of prosthetic material. Split thickness skin grafts can be placed directly on the omentum. This technique is frequently less than ideal cosmetically, and is generally used when other flaps are either unavailable or have been unsuccessful.

Free vascularized tissue flaps, created using microsurgical techniques, have recently been used to provide soft tissue coverage of chest wall defects. The tensor fascia lata free flap, which consists of the tensor fascia lata muscle, its overlying fascia and skin, is based on the lateral femoral circumflex vessels. These vessels are anastomosed to either the thoracoacromial or transverse cervical vessels. It is also possible to use vascularized fibular bone grafts to replace missing ribs.

Any tissue being transferred must be well vascularized. Particular consideration must be given to radiation for carcinoma of the breast which is directed to the internal mammary and axillary nodes; these are adjacent to the vascular pedicles of pectoralis and latissimus flaps, respectively. Although prior irradiation does not prohibit use of that flap, preoperative arteriography may be useful in assessing the adequacy of the blood supply.

Adkins PC. Tumors of the chest wall. In: Shields TW, ed. General Thoracic Surgery, 2nd edn. Philadelphia: Lea and Febiger, 1983: 453–63.

Bostwick J, Scheflan M, Nahai F, et al. The ‘reverse’ latissimus dorsi muscle and musculocutaneous flap: anatomical and clinical considerations. Plast Reconstr Surg 1980; 65: 395–399.

Boyd AD, et al. Immediate reconstruction of full thickness chest wall defects. Ann Thorac Surg 1981; 32: 337–46.

Cavanaugh DG, Cabellon S, Peake JB. A logical approach to chest wall neoplasms. Ann Thorac Surg 1986; 41: 436–7.

Chang AE, Rosenberg SA, Glatstein EL, Antman KH. Sarcomas of soft tissue. In: DeVita VT, Hellman S, Rosenberg SA, eds. Cancer: Principles and Practice of Oncology, 3rd edn. Philadelphia: JB Lippincott, 1989: 1244–92.

Enzinger FM, Weiss SW. Soft Tissue Tumors, 2nd edn. St Louis: CV Mosby Co. 1988: 122–7, 201–22.

Graeber GM. et al. Initial and long-term results in the management of primary chest wall neoplasms. Ann Thorac Surg 1982; 34: 664–73.

Graham J, Usher FC, Perry JL, Barkley HT. Marlex mesh as a prosthesis in the repair of thoracic wall defects. Ann Surg 1960; 151: 469.

King RM, Pairolero PC, Trastek VF, Piehler JM, Payne WS, Bernatz PE. Primary chest wall tumors: factors affecting survival. Ann Thorac Surg 1986; 41: 597–601.

McAfes MK, et al. Chondrosarcoma of the chest wall: factors affecting survival. Ann Thorac Surg 1985; 40: 535–41.

McKenna RJ, Mountain CF, McMurtrey MJ, Larson D, Stiles QR. Current techniques for chest wall reconstruction: expanded possibilities for treatment. Ann Thorac Surg 1988; 46: 508–12.

Marin ML, Perzin KH, Markowitz AM. Elastofibroma dorsi: benign chest wall tumor. J Thorac Cardiovasc Surg 1989; 98: 234–8.

Melawar MM, Link MP, Donaldson SS. Sarcomas of bone. In: DeVita VT, Hellman S, Rosenberg SA, eds. Cancer: Principles and Practice of Oncology, 3rd edn. 1989. Philadelphia: JB Lippincott, 1293–1342.

Morgan RF, Edgerton MT, Wanebo HJ, Daniel TM, Spotnitz WD, Kron IL. Reconstruction of full thickness chest wall defects. Ann Surg 1988; 207: 707–16.

Pairolero PC. Chest wall tumors. In: Shields TW, ed. General Thoracic Surgery, 3rd edn. Philadelphia: Lea and Febiger, 1989.

Pairolero PC, Arnold AG. Chest wall tumors—experience with 100 consecutive patients. J Thorac Cardiovasc Surg 1985; 90: 367–72.

Pass HI. Primary and metastatic chest wall tumors. In: Roth JA, Ruckdeschel JC, Weisenburger TH, eds. Thoracic Oncology, Philadelphia: WB Saunders, 1989; 546–65.

Rami-Porta R, Bravo-Bravo JL, Aroca-Gonzalez MJ, Alix-Treuba A, Surrano-Munoz F. Tumours and pseudotumours of the chest wall. Scand J Thorac Cardiovasc Surg 1985; 19: 97–103.

Sabanatham S, Salama FD, Morgan WE, Harvey JA. Primary chest wall tumors. Ann Thorac Surg 1985; 39: 4–15.

Thomas PRM, et al. Primary Ewing's sarcoma of the ribs. Cancer 1983; 51: 1021–7.