ELSEt’lER Radiotherapy and Oncology 34 ( 1995) y-16

R ADIOTHERAPV

4S ONCOLOGY

Thymoma: results of a multicentric retrospective series of 149 non- metastatic irradiated patients and review of the literature

D. Cowen*, P. Richaud, F. Mornex, T. Bachelot, G.M. Jung, X. Mirabel, C. Marchal, J.-L. Lagrange, P. Rambert, G. Chaplain, T.D. N’Guyen, M. Resbeut,

for the FNCLCC trialists

Received 2 June 1994: revision received 18 October 1994: accepted 25 Nocember 1994

Abstract

Between 1979 and 1990, 149 patients with non-metastatic thymomas were treated in ten French cancer centers. Patients were staged according to the ‘GETT’ classification, derived from that of Masaoka. There were 13 stage I patients. 66 stage 11. 58 stage III and 32 stage IVA. Gross total resections were performed in 63 cases, subtotal resections in 31 cases and in _(5 cases a biops) alone was performed. All patients received radiotherapy and 74 were given post-operative chemotherapy. Median follow-up was 7.7 years. Local control was achieved in 117 cases (78.5%) and was influenced by the stage of the disease @ < 0.01) and the extent of surgery (p < 0.01). Twenty-six patients developed metastatis after a median period of 9 months. Five- and ten-year disease-free survival rates were 59.5% (Sl-67”X) and 49.5% (39-60X), respectively. and were influenced by the stage of the disease (p < 0.01 L the extent of surgery (p c 0.001) and a mediastinal compression on presentation (p = 5 x IO-‘). Four factors could predict a worse overall survival in the multivariate analysis: mediastinal compression on presentation (p < O.OOl), absence of chemotherap) (p < O.OOl), biopsy alone @ = 0.003), and young age @ = 0.013). A worse DFS was predicted by mediastinal compression on presen- tation @ c O.OOl), absence of chemotherapy @ < O.OOl), young age (p = 0.006). and stages Ill-IVA (p = 0.04). Future therapeutic strategies are discussed and the literature is reviewed.

Keywords: Thymoma; Radiotherapy; Surgery; Chemotherapy; Prognostic factors

1. Introduction

Thymomas are rare, usually slow-growing, epithelial neoplasms, considered as malignant because of their potential invasiveness. Invasion beyond the capsule is correlated with a worse outcome if they were treated by surgery alone, although the appearance of the cells under the microscope suggests they are benign. The dis- crepancy between the histological appearance and the capability of loco-regional extension explains why reports have used the term ‘malignant thymoma’ to designate invasive thymomas [26]. The staging and opti- mum treatment of thymomas remain controversial. Although surgery remains the treatment of choice,

* Corresponding author, Department of Radiotherapy. Institut Paoli-Calmettes. 232 Bvd. Sainte Marguerite. F- I3009 Marseille. France.

radiotherapy is widely used in invasive and. for some. also in non-invasive thymomas. The place of chemother- apy is being evaluated in prospective trials. but the answers could take a long time to emerge because of the low incidence of the disease and the necessary long follow-up. To define the most appropriate therapeutic strategies we report a large multicentric retrospcctlve series of 149 patients treated in ten French cancer centers, and a literature review is presented.

2. Materials and methods

2.1. Chical feutures

From 1979 to 1990. 149 patients uere refered to Icn French cancer centers for the treatment of d non- metastatic thymoma. During the same period. only 1-I other patients had a metastatic disease on presentation.

Ol67-8140/95/$09.50 B 1995 Elsevier Sclrnce Ireland Ltd. All rights rcwr\ed SSDI 0167-8140(94)01493-M

10 D. Cowm et rd. / Rodiorhrrupy und Oncology 34 (1995,! 9-16

Table I Means of discovery on presentation

No. of patients (‘l/u)

Chest pain, cough, dyspnea Myasthenia gravis Superior vena caval compression Abnormal chest X-ray Constitutional symptoms Erythroblastopenia Other

33 (22) 32 (21) 31 (21) 29 (19) 16 (II) I (0.6) I (5)

The median follow-up was 92 months with a range of 17-164 months. The age ranged from 15 to 81 years with a median of 45 years. The ratio of female:male was 1.1: 1. Table 1 summarizes the means of discovery of the thymoma. Mediastinal compression concerned 87 patients and was defined as at least one of the following symptoms: superior vena caval compression, chest pain, bronchial compression, dyspnea and pleural effusion,

2.2. Pathologic classification

All pathologic specimens and operative reports were reviewed in each center and a centralized re-review was performed for all cases which seemed uncertain to the pathologist. The diagnosis of thymoma was based on the pathologic classification of Rosai and Levine [36], that is, neoplasms of thymic epithelial cells, regardless of the presence of a lymphoid component. The definition was modified [25] by the same authors, such that thymic tumours containing cells with the cytological aspect of malignancy are separately classified as thymic car- cinomas. Therefore, such carcinomas which have a very different clinical outcome [26,34] were excluded from this study. We also excluded neurogenic tumours, non- Hodgkin’s and Hodgkin’s lymphomas. The tumours were classified into lymphocytic type (n = 18), epithelial type (n = 60) or mixed lymphoepithelial type (n = 71), as described by Bernatz et al. (51.

2.3. Staging

Staging was based on the surgical and pathologic fea- tures described by the French Study Group on Thymic Tumours (GETT) [ 13,171 derived from the Masaoka staging system [30] (Table 2). In the GETT staging sys- tem (Table 2), the predominant feature is the extent of surgical resection: non-invasive completely resected tumours are stage I and invasive completely resected tumours are stage II tumours, whereas incompletely resected tumours are stage III tumours. In the Masaoka staging, a contiguous tumour spread to the pericardium is a stage III tumour, whereas in the GETT system, in

Table 2 GETT and Masaoka staging systems

GETT stuging Stage I

I-A Encapsulated tumour, totaly resected I-B Macroscopically encapsulated tumour, totaly resected,

but the surgeon suspects mediastinal adhesions and potential capsular invasion

Stage II Invasive tumour, totaly resected Stage III

III-A Invasive tumour, subtotaly resected III-B Invasive tumour, biopsy

Stage IV IV-A Supraclavicular metastasis or distant pleural implant IV-B Distant metastasis

Masaoka staging Stage I Macroscopically completely encapsulated and no micro-

scopic capsular invasion Stage II

I Macroscopic invasion into surrounding fatty tissue or mediastinal pleura

2 Microscopic invasion into the capsule Stage III Macroscopic invasion into neighboring organs Stage IV

IV-A Pleural or pericardial implants IV-B Lymphogenous or hematogenous metastasis

case of total resection, it becomes a stage II tumour. Reciprocally, an encapsulated totaly resected stage I (GETT) thymoma becomes a stage II-2 of Masaoka in case of a microscopic invasion of the capsula. The cor- respondence between the Masaoka and the GETT sys- tem was studied and the two systems appeared concordant in 88% of cases as summarized in Table 3. For staging work-up, CT of the chest, lung function and blood tests were included routinely. Bone scan was in- cluded after 1985 in all centers.

2.4. Surgery

Extent of resection was determined by operative reports. Sixty-three patients underwent gross total resec-

Table 3 Correspondence between GETT and Masaoka staging systems

GET-T Masaoka Total

1 II 1 II 2 III IVA

IA 1st I 6 IB 7 II 9 46 III A III B I IV A

Total 5 29 I7 66 32 149

D. Cowen et al. /Radiotherapy and Oncolog,y 34 (1995) 9-16 II

Table 4 Doses of radiotherapy as a function of stage and extent of surgery

GETI stage Median dose delivered bw)

I (n = 13) 40 Gy (35-50) II (n = 46) 50 Gy (38-60) III-A (n = 21) 54 Gy (38-70) III-B (n = 37) 50 Gy (33-65) IV-A (n = 32) 50 Gy (38-68)

Exient qf surgery Total resection (n = 63) Subtotal resection (n = 31) Biopsy alone (n = 55)

50 Gy (35-68) 55 Gy (38-70) 50 Gy (30-65)

tion (all stages I and II but also, according to operative reports, 4 stages IVA), whereas 31 patients (10 stage IVA and 21 stage III-A) underwent subtotal resection (> 50% and < 100% resected). Fifty-five tumours were not resected and a biopsy alone (< 50% resected) was performed (stage III-B in 37 cases and IVA in 18 cases).

2.5. Radiotherapy

All the patients received radiation therapy, using high energy photon beams (> 10 MeV). The volume treated included the entire mediastinum and part of the adjacent lung when there was a parenchymal involvement or as delineated by CT scan or surgical clips, plus a 2-cm margin. Patients received 2 Gy fractions through AP-PA equally weighted portals. Anterior wedged portals were used when an additional boost was given and when the dose to the spinal cord had reached 40-44 Gy. In 97 cases, both supraclavicular fossae were also treated and a median dose of 40 Gy was delivered. The doses of radi- ation delivered as a function of stage and of extent of surgery are presented in Table 4. usually post-operative except for received pre-operative (22-50 Gy) (22-35 Gy) irradiation.

2.6. Chemotherapy

Radiotherapy was five patients who and post-operative

The use of chemotherapy was strongly correlated with stages III and IVA (p < O.Ol), and with mediastinal compression on presentation (J < 0.01). In 51 cases the multi-agent platinum-based CAP regimen was used [ 111. Chemotherapy always followed surgery but either preceeded (n = 24) or followed radiotherapy (n = 43) or both (n = 7). Chemotherapy and radiotherapy were never delivered concomitantly. The administration of chemotherapy as a function of stage is presented in Table 5. Seventy-four patients (50%) were treated with chemotherapy as a part of the treatment.

Table 5 Chemotherapy as a function of stage

GETT Stage Chemotherapy (% patients)

I (n = 13) 5/13 (38) II (n = 46) IO/46 (22) III-A (n = 21) 12/21 (57) III-B (n = 37) 24137 (65) IV-A (n = 32) 23/32 (71)

2.7. Statistics

The overall survival (OS) was calculated from the date of surgery until death of any cause. Disease-free survival (DFS) was calculated from the date of diagnosis until evidence of recurrence or progression. Survival curves were calculated using the Kaplan and Meier non- parametric estimation [20], and differences were tested using the log-rank test. Confidence intervals were calculated according to Rothman’s method [22]. The multivariate analysis used the stepwise Cox model of the Biomedical Programs (BMDP) package from University of California (Los Angeles, CA) [lo].

3. Results

3.1. Local control

Local control was defined as sustained progression- free interval. With a median follow-up of 7.7 years (range, 1.4-13.7 years), local control was achieved in 117 cases (78.5%). Among the 32 remaining patients, 20 never achieved a complete remission (the patients pro- gressed), and in 12 cases a local relapse was observed after a median period of 19 months (range, 12-66 months). Local relapse and progressive disease occured within the irradiated fields in 26 cases (8 1.2%) and were out-field relapses in six cases (18.8%). The occurence of local relapse as a function of stage is presented in Table 6. The rate of local failure was 2% (l/59) in stages I and

Table 6 Local relapse as a function of stage

GETT Stage n Progressive Local Local failures disease relapse total (%)

IA 6 -- - 0 IB 7 - - 0 II 46 - 1 1 (2%) III A 21 2 I 3 (14%) III B 37 9 6 15 (46%) IV A 32 9 4 13 (41%)

Total 149 20 12 32 (21%)

12 D. Cowen et al. /Radiotherapy and Oncology 34 (1995) 9-16

II (totally resected) versus 34.4% (31190) in stages III and IVA (p < 0.01). The extent of surgery influenced local control. We observed 2/63 (3%) failures after total resection, 5/3 1 (16%) after subtotal resection and 25/55 (45%) after biopsy alone @ < 0.001, overall). No rela- tionship was found between the dose of irradiation and the occurence of a local failure: the median dose delivered to patients who failed was 51.8 Gy versus 50.5 Gy for those controlled. This was also true when only stage III and IVA were considered: in such cases we observed 15/33 (45.5%) recurrences after delivering doses lower than 50 Gy, and 16/57 (28.1%) recurrences after delivering doses above 50 Gy @ = 0.1). Finally, no relationship was found between the use of chemother- apy and the probability of achieving local control, in the univariate analysis.

3.2. Metastasis

Twenty-six patients (17%) presented a metastatic spread after a median period of 9 months (range, 2-76 months), associated with a local relapse in three cases and with progressive disease in six cases. The sites of metastatic spread were the lung (n = 9), liver (n = 9) bone (n = 3), adrenal glands (n = 3), kidney (n = 3), neck nodes (n = 2), skin (n = l), central nervous system (n = 1). Metastatic spread was less frequent in patients who underwent gross total resection (7.9%) than in those who underwent a subtotal resection (22.6%) or a biopsy alone (27.3%). The difference is statistically signi- ficant (p < 0.02). Because a correlation of stage with metastatic spread is probably more relevant than extent of resection, and because the GETT classification is very strongly correlated with the extent of resection, we look- ed at the influence of the Masaoka stage on the oc- curence of metastasis. We observed 015, 3146, 15166, and 8/32 metastasis, respectively, in stages I, II, III and IVA (p = 0.1). The rate of metastasis was reduced by chemo- therapy in stages III-IVA: we observed lo/58 and 12/32 metastasis, respectively according to the use of chemo- therapy or not Cp < 0.05). There was a trend towards a reduction of the rate of metastasis in stages I-II and we observed l/14 and 3/41 metastasis, respectively accord- ing to the use of chemotherapy or not (p = 0.1).

3.3. Causes of death, and survival

Sixty-two patients died. In 47 cases, death was tumour-related (14 progressive diseases, 9 local relapses, 15 metastasis, and all three causes in 9 cases), whereas 14 patients died of intercurrent disease and one patient died of non-Hodgkin’s lymphoma after having been treated with 24 courses of chemotherapy. Actuarial 5- year DFS according to the stage of disease appears in Fig. 1. The DFS rate was 92% (confidence interval (CI), 67-99%) for stage I, 75% (CI, 61-86%) for stage II, 60%

100%

75%

50%

25% I . .

p<O.Ol ) OVeralJ 0% +

0 5 10 years

Fig. I. Disease-free survival according to the stage of disease.

(CI, 38-78%) for stage III-A, 39% (CI, 25-56%) for stage III-B, and 48% (CI, 32-66%) for stage IVA. Overall, there was a statistically significant difference by stage (p < 0.01). The same statistically significant differ- ence by stage was observed for overall survival.

The 5-year OS rate was 92% (CI, 67-99%) for stage I, 77% (CI, 62-87%) for stage II, 67% (CI, 45-82%) for stage III-A, 41% (CI, 27-58%) for stage III-B, and 51% (CI, 34-68%) for stage IVA. When patients with non- invasive thymoma (Stage I) were compared with those with invasive thymoma (Stage II-IVA), the 5-year DFS rate was 92% (CI, 66-98%) and 57% (CI, 48-65%), respectively (p = 0.07).

Fig. 2 shows the DFS by the extent of surgery. The 5- year DFS rate was 74% (CI, 60-840/o) for the patients who underwent total resection, 60% (CI, 40-77%) for subtotal resection, and 38% (CI, 26-51%) for biopsy alone (p < 0.001, overall). The respective 5-year OS figures were 80% (CI, 68-88%), 64% (CI, 44-80%) and 39% (CI, 27-52%). When patients who had total resec- tion were subdivided into those having non-invasive and invasive thymoma, the 5-year DFS rates were not statistically different (92% and 75% respectively, p = 0.50). Finally, in the univariate analysis, patients with a mediastinal compression on presentation ex- perienced a worse outcome than without. The IO-year DFS rate was 25% in case of mediastinal compression

100%

75%

50%

25%

0%

,

Biqmydone p<O.ool, ovemu

0 5 10 years

Fig. 2. Disease-free survival according to the extent of surgery.

D. Cowen et al. /Radiotherapy and Oncology 34 (1995) 9-16 13

100%

75%

50%

25%

0% 5 10 years

Fig. 3. Disease-free survival of stage III-IV patients with a mediastinal compression on presentation according to the administra- tion of chemotherapy (CT) or not.

on presentation and 77% without (p = 5 x 10v6). Myasthenia gravis was observed in 32 cases, of which none was associated with a mediastinal compression.

A multivariate analysis was performed to determine which factors could predict disease-free survival and overall survival. The following potential risk factors which might have influenced the occurence of death or relapse were investigated in the Cox model: age, chemo- therapy, stage, mediastinal compression on presenta- tion, myasthenia gravis, and extent of surgery. Four of these wereshown to have a statistically significant influ- ence on the occurence of a worse DFS: mediastinal com- pression on presentation (p < O.OOl), no chemotherapy @ < O.OOl), age less than 30 years @ = 0.006) and stages III-IVA @ = 0.04). The beneficial effect of chemother- apy is shown in Fig. 3 which reports the DFS for stage III-IVA patients with mediastinal compression on pres- entation according to the administration of chemother- apy or not. Four factors were also shown to adversely affect overall survival: mediastinal compression on pres- entation @ < O.OOl), no chemotherapy (p < O.OOl), bi- opsy alone 0, = 0.003), and age less than 30 years @ = 0.013).

3.4. Complications

Seven patients developed grade 3 or 4 (WHO grading) complications. Three pericarditis occured after radia- tion doses of 36, 55 and 60 Gy, and four patients developed lung fibrosis after receiving lung doses of 40, 54, 62 and 65 Gy. One patient experienced a probably treatment-induced non-Hodgkin’s lymphoma after hav- ing received 24 courses of chemotherapy.

4. Discussion

4.1, Myasthenia gravis

In older reports [23,37,41], myasthenia gravis (MG) seemed to adversely affect survival because of a high in-

cidence of per&operative deaths. Medical treatments have evolved and such deaths can be avoided nowadays. In 1966, Wilkins et al. [41] concluded that the principal factors related to survival were ‘the invasive nature of the tumour and the presence of MG’. However, the same authors reported more recently that the last MG- related death after thymectomy was in 1963 [42], like the last MG-related death reported by Jackson and Ball in 1991 [ 191. Our data are in accordance with the opinion of many recent papers which eliminated MG as a signifi- cant factor in long-term survival [ 13,19,26,30,34, 38,391. Moreover, Maggi et al. reported a favourable outcome because of earlier detection of the thymoma [28], whereas Nakahara et al. reported higher rates of resectability and a higher incidence of stages I in case of MG [33], and Monden et al. found fewer local failures for equal stages [31], when MG was present.

4.2. Pathologic classification

The prognostic significance of pathologic classifica- tion is controversial. Histological aspects of non- invasive and invasive tumours can be identical [ 14,16, 29,401, and the division into lymphocytic, epithelial or both types remains subjective. The supposed influence of histological subtypes may disappear when it is linked with strong prognostic variables such as the stage or the extent of surgical excision [26,33]. However, some reports demonstrated a poor outcome of epithelial [3 1,391 or lympho-epithelial subtypes [34], whereas others associated lymphocytic subtypes with a favour- able outcome [5,27,31] as well as spindle-shaped or oval cell subtypes. In this report, the pathologic classification had no influence on outcome.

4.3. Staging

Tumour invasion beyond the capsule is associated with an unfavourable outcome [5,9,12,18,28,30,33,34, 38,391 and should be considered as the first means of classification: in case of non-invasive thymoma, the relapse rate is very low [28] or nil1 [9,12,18] as in the pre- sent report. In an extensive review, Momex et al. [32] reported an average 80% survival for non-invasive thymomas versus less than 50% in case of invasion. The classifications of Bergh et al. [4], Wilkins et al. [41], Masaoka et al. [30], Curran et al. [9], Verley and Hollmann [39], and the GETT classification [ 171 con- sider the extent of capsular invasion. In three classifica- tions [4,9,41] a capsular invasion with no extension into the mediastinal fat is considered as a stage I: in such cases Curran et al. [9] showed that the patients shared the prognosis of those with no capsular invasion. On the other hand, Masaoka et al. [30] pooled microscopic cap- sular invasions with extension to the mediastinal fat or pleura into stage II, whereas invasion to pericardium

14 D. Cowen et al. /Radiotherapy and Oncology 34 (1995) 9-16

was considered as stage III. Among invasive turnouts, survival is usually worse when the stage is higher: stage III tumours have a better prognosis than stage IVA tumours [28,33,38]. In the present report, contrary to several others [9,30,34,39], the differences of DFS observed are statistically significant overall (p < 0.01). Altogether, non-invasive thymomas have the best prog- nosis and stage IVA the poorest. For ‘transition’ stages, the extent of surgery seems the predominant prognostic factor.

4.4. Extent of surgery

When possible, the first line of therapy for thymoma should be surgery and all published reports bear higher overall survival rates and DFS in case of total resection [9,24,37], or when compared with partial excision [ 13,26,30,33,35] although the differences observed are not always statistically significant [28,34,38]: 5-year sur- vival range from 74-98% in the case of total resection [32]. Progress in the field of vascular surgery might help to enhance the number of total resections, especially in case of associated vena caval obstruction [33]. Concern- ing subtotal surgery, the analysis of published reports is misleading: the use of radiotherapy differs from one report to another, the extent of surgery is not always precisely described [26], the number of patients con- cerned is often small, and the studies were performed over periods of time up to 40 years. Subtotal resections seem to produce higher survival rates than simple biop- sies [28,33,34]. However, the number of patients who underwent a biopsy is small - lo-20% - even in the series reporting more than 100 patients [9,26,28,33, 35,391 and the data should therefore be analysed cautiously. Five-year survival rates range from 64% to 71% after subtotal resection, compared with 25-40% in case of biopsy alone (Ref. [28,33] and present report). The advantage of partial surgery over biopsy alone re- mains a controversial point for several authors [2,9,13]: in the present report, and as in others [34,38], we observed more metastasis and more local recurrences after simple biopsies than after subtotal resections.

4.5. Radiotherapy

There is a general agreement about the relevance of post-operative radiotherapy after subtotal resection or biopsy alone [2,9,19,21,28,32,34,38]. The literature published on this topic reports series of 1 l-50 patients with local control rates of 50-80% [2,9,19,21,34,38], and more local failures among patients with biopsies [38]. To reduce the rate of out-field failures, the use of pro- phylactic pleural and pulmonary irradiation has been advocated [18,38], as well as the use of radioactive 32P [3]. This seems questionable in consideration of the low doses delivered. Furthermore, although out-field recur-

rences seem more frequent in stage IVA diseases, they are often associated with in-field mediastinal recur- rences. Due to the smallness of the samples reported in the literature, no conclusions can be drawn about the usefulness of preoperative radiotherapy (Refs. [28,35] and present report). We observed one third of local fail- ures after subtotal resections or simple biopsies, of which 61% were in-field. Local control was achieved in one half of the patients after biopsy alone. Chemother- apy did not influence local control and our results were obtained with median doses of 50 Gy. In accordance with others [2,19] this could suggest that there was no relationship between the dose of radiation and local relapses although pooled data [8] suggested doses of 60 Gy or more could improve local control. A field of fu- ture investigations to enhance local control with radiotherapy could be the use of intraoperative radiotherapy due to its ability of delivering a high dose in a reduced volume, as well as conformal radiotherapy due to the advent of 3D imaging procedures and their accuracy in delineating the target volumes. These de- vices could also help to avoid radiation-related com- plications: although unusual in our series (4.5%) such complications are difficult to avoid with high energy photon beams because of the closeness of the heart and the lungs. Therefore, we recommend neither to employ fraction sizes exceeding 2 Gy, nor cardiotoxic chemo- therapy.

The relevance of radiotherapy after total resection is more controversial and should be considered according to the invasiveness of the thymoma. In a review of eight reports concerning 115 invasive tumours completely resected, Curran et al. reported a 5% recurrence rate with the use of postoperative irradiation versus 28% without radiotherapy [9]. Nakahara et al. reported the same outcome for non-invasive completely resected thymomas than for invasive completely resected and ir- radiated tumours [33]. Although this point remains questionable [7,38] our experience provides additional data favouring the use of postoperative radiotherapy with equal survivals for stages I and II. According to the literature and our results, a dose of 50 Gy should be suf- ficient in such cases [ 1,9,34,39] although Nakahara et al. [33] suggested 40 Gy could achieve a 93% local control rate for completely resected stage II and III diseases. Regarding non-invasive completely resected thymomas, the analysis of published literature does not favour the use of adjuvant radiotherapy. In such a situation, Fu- jimora et al. and Curran et al. did not experience any local failures [9,12], whereas Nakahara et al., since 1988, were currently using surgical treatment alone for stage I disease [33]. However, Pollack et al. believed that an exception could be made for large stage I disease in case of suspicion of pleural or pericardial adhesions (GETT- stage IB) during surgery [34]. An adequate microscopic examination for sites of discontinuity through the cap-

D. Cowen et al. /Radiotherapy und Oncolog!~ 34 ( IYYS) Y-16 IS

sule, and microscopic invasion into adherent pleura or pericardium is often impossible because of the thinness of the capsule and its ability of being broached during handling or preparation [34]. Since no failures among ir- radiated stage IB patients have been observed in our series, we share this point of view and advocate the use of postoperative radiotherapy in those stage I patients.

4.6. Chemotherapy

The role of chemotherapy still requires prospective evaluation (71. Prednisone has been used because of its thymolytic effects on cortical lymphocytes [ 151, often in association with cisplatinum [21]. Complete response rates of 43% were obtained by using a regimen associating cyclophosphamide, adriamycin and cispla- tinum (CAP) [ 111, and the Radiation Therapy Oncology Group has recently initiated a trial testing the use of neoadjuvant CAP prior to radiotherapy for unresectable thymoma [34]. Unfortunately, the study closed because of poor accrual. In the present report, the multivariate analysis showed that the use of chemotherapy was associated with higher survival rates. This favourable outcome concerned patients with stage III-IVA disease and mediastinal compression on presentation (Fig. 3) which were independent factors predicting an un- favourable outcome in the multivariate analysis. Still, it is too early to advocate the use of adjuvant chemother- apy in advanced thymomas, and prospective trials are needed.

5. Conclusion

Surgery remains the main treatment of invasive and non-invasive thymomas. We advocate the use of post- operative radiotherapy delivering doses as high as 50 Gy for GE’IT stage IB thymomas as for invasive completely resected thymomas. When a simple biopsy or an incom- plete resection is performed, local control can still reach 65%. In.such cases, we suggest new treatment modalities such as extensive surgery, preoperative or intraoperative radiotherapy could be evaluated. Last, 25% of stage III and IVA thymomas develop metastasis and the use of adjwant chemotherapy requires prospective evaluation.

References

111 Arakawa, A., Yasunaga, T., Saitoh, Y., Uozumi, H., Takada, C., Baba, Y., Yoshizumi, K. and Takahashi, M. Radiation ther- apy of invasive thymoma. Int. J. Radiat. Oncol., Biol. Phys. IS: 529-534, 1990.

[2] Arriagada, R., Bretel, J.J., Caillaud, J.M., Garreta, L., Guerin. R.A., Laugier, A., Le Chevalier, T., and Schlienger, M. Invasive carcinoma of the thymus. A multicenter retrospective review of 56 cases. Eur. J. Clin. Oncol. 20: 69-74, 1984.

]3] Batata, M., Martini, N., Huvos, A.G., Aguilar. R.I. and Beattie, E.J. Jr. Thymomas: clinicopathologic features. therapy and prognosis. Cancer 34: 389-396, 1974.

[4] Bergh, N.P., Gatzinsky. P.. Larsson. S.. Lundin. P. and Rindell. B. Tumor of the thymus and thymic region. I. Clinicopatho- logical studies on thymoma. Ann. Thorac. Surg. 25: 91-98. 1978.

IS] Bernatz, P.E., Khonsari, S., Harrison, E.G. and Taylor. W.F. Thymoma: factors influencing prognosis. Surg. Clin. North Am. 53: 885-893, 1973.

[6] Chahinian, A.P., Bhardwaj, S., Meyer, R.J.. Jaffrey. I.S.. Kirschner, P.A. and Holland, J.F. Treatment of invasive or metastatic thymoma. Report of I I cases. Cancer 47: 1752- 1761. 1981.

171 Cohen, D.J., Ronnigen, L.D., Graeber, G.M., Deshong. J-L.. Jaffin, J.. Burge, J.R. and Zajtchuk, R. Management of patients with malignant thymoma. J. Thorac. Cardiovasc. Surg. 87: 301-307. 1984.

181 Cox. J.D. (1989) The lung and thymus. In: Radiation Oncology Rationale, Technique, and Results. pp. 305-308. Editors: W.T. Moss and J.D. Cox. C.V. Mosby Company, St. Louis.

[9] Curran, W.J.. Kornstein. M.J.. Brooks. J.J. and Turrisi, A.T. Invasive thymoma: the role of mediastinal irradiation following complete or incomplete surgical resection. J. Clin. Oncol. 6: 1722-1727, 1988.

[IO] Dixon W.J. BMDP Statistical Software. Berkeley, University of California, 1983. pp. 330-344.

[II] Fornasiero, A., Daniele. 0.. Ghiotto. C.. Piazza M.. Fiore- Donati, L., Calabro. F.. Rea. F. and Fiorentino, M.V. Chemo- therapy for invasive thymoma: a 13 year experience. Cancer 68: 30-33, 1991.

]l2] Fujimura, S.. Kondo, T., Honda. M.. Shiraishi, Y.. Tamahashi, N. and Nakada, T. Results of surgical treatment of thymoma based on 66 patients. J. Thorac. Cardiovasc. Surg. 93: 708-714. 198?.

[I9 Gamondbs, J.P., Balawi, A., Greenland, T., Adleine, P.. Mornex, J.F.. Zhang, J. and Maret. G. Seventeen year< of surgi- cal treatment of thymoma: factors influencing survival. Eur. J. Cardio-thorac. Surg. 5: 124- I3 I. 1991.

[I41 Gerein, A.N., Srivastava. S.P. and Burgess. J. Thymoma: a ten year review. Am. J. Surg. 136: 49-53, 1978.

[I51 Gonzalez Gonzalez, D. The need for prospective clinical studies in thymomas. Radiother. Oncol. 21: 75-76. 1991.

[I61 Gray, G.F. and Gutowski, W.T. Thymoma: A clinicopathologic study of 54 cases. Am. J. Surg. Pathol. 3: 235-249. 1979.

[17] Gutrin. R.A., Malhaire, J.P.. Touboul. E.. Catach. E. and Richard, P. Radiotherapy of lymphoepithelial thymomas. Re- sults in 29 cases. J. Eur. Radiother. 5: 309-312, 1984.

[18] Haniuda, M., Morimoto. M.. Nishimura. H.. Kobayashi. 0.. Yamanda, T. and Lida, F. Adjuvant radiotherapy after com- plete resection of thymoma. Ann. Thorac. Surg. 54: 31 I-31, 1992.

[I91 Jackson, M.A. and Ball, D.L. Post-operative radiotherapy in in- vasive thymoma. Radiother. Oncol. 21: 77-82. 1991.

[20] Kaplan, E.L. and Meier. P. Non parametric estimation from in- complete observations. J. Am. Stat. Assoc. 53: 457-481. 1958.

(211 Krueger, J.B., Sagerman. R.H. and King. G.A. Stage Ill thymoma: results of postoperative radiation therapy. Radiology 168: 855-858, 1988.

(221 Laplanche, A., Corn-Nouque, C. and Flamant. R. Methodes Statistiques Appliquees a la Recherche Clinique. Flammarion. Paris 1987.

(231 Legg, M.A. and Brady. W.J. Pathology and clinical behaviour of thymomas: a survey of 51 cases. Cancer 18: I I3 I - I 144. 1965.

[24] Le Golvan. D.P. and Abell. M.R. Thymomas. Cancer 39: 2142-2157. 1977.

[25] Levine. G.D. and Rosai. J. Thymic hyperplasia and neoplasia: a review of current concepts. Hum. Pathol. 9: 495-515. 1978.

[26] Lewis, J.E.. Wick. M.R.. Scheithauer. B.W.. Bernatz. P.E. and

Taylor. W.F. Thymoma: a chnrcopathologic revrew Cancer 60: ation in 141 consecutive patients. J. Thorac. Cardiovasc. Surg. 2721-2743. 1987. 95: 1041-1047. 1988.

[27] Maggi. G.. Graccone. G.. Donadio. M.. Ciuffreda. L.. Daleslo. 0.. Leria. G.. Trifilettr. G.. Casadio. C.. Palestro. G.. Mancuso. M. and Calciati. A. Thymomas: a review of 169 cases. with par- ticular reference to results of surgical treatment. Cancer 58: 765-716. 1986.

1341

[351

[28] Maggi. G.. Casadio. C.. Cavallo. A.. Ciancr. R.. Molinatti. M. and Rufftni. E. Thymoma: results of 241 operated cases. Ann. Thorac. Surg. 51: 152-156, 1991. [361

[29] Marchevsky. A.M. and Kaneko. M. Surgical Pathology of the Mediastinum. Raven Press. New York. 1984.

[30] Masaoka. A.. Monden. Y.. Nakahara, K. and Tanioka. T. Follow-up study of thymomas with special reference to their clinical stages. Cancer 48: 2485-2492. I98 I.

[31] Monden. Y.. Nakahara. K.. lioka. S.. Nanjo. S.. Ohno, K.. Fujii. Y.. Hashimoto. J.. Kitagawa. Y.. Masaoka, A. and Kawashima. Y. Recurrence of thymoma: clinico-pathological features. therapy and prognosis. Ann. Thorac. Surg. 39: 165-169. 1985.

1321 Mornex. F.. Resbeut, M.. Richaud. P.. Jung, G M.. Mirabel. X.. Marchal. C.. Lagrange. J.L.. Rambert. P.. Chaplain, G. and Nguyen. T.D. Invasive thymomas: the role of radiotherapy following partial surgical resection or biopsy. A multicentric ret- rospective review of 90 cases. Int. J. Radiat. Oncol.. Biol. Phys. (in press).

[371

[381

1391

[401

1411

1421 1331 Nakahara. K.. Ohno. K.. Hashimoto. J.. Maeda. H.. Miyoshi.

S.. Sakurai. M.. Monden. Y. and Kawashima. Y. Thymoma: re- sults with complete resection and adjuvant postoperative irradi-

Pollack. A., Komaki. R.. Cox. J.D.. Ro. J.Y.. Oswald, M.J.. Shin. D.M. and Putnam, J.B. Thymoma: treatment and progno- sis. Int. J. Radiat. Oncol.. Biol. Phys. 23: 1037-1043. 1992. Ribet. C.. Voisin. C.. Go&in, B., Pruvot. F.R., Ramon, P., Wallaert. B. and Dambron. P. Les thymomes lympho- Cpithlliaux: etude anatomo-clinique et therapeutique de I 13 cas. Rev. Mal. Resp. 5: 53-60, 1988. Rosai. J. and Levine. G. Tumors of the thymus. In: Atlas of Tumor Pathology. Series 2. Fascicle I3 (Armed Forces Institute of Pathology) Washington DC. 55-99. 1976. Salyer. W.R. and Egglestone. J.C. Thymoma: a clinical and pathological study of 65 cases. Cancer 37: 229-249. 1976. Urgesi. A.. Monetti, U.. Rossi, G.. Ricardi. U. and Casadio. C. Role of radiation therapy in locally advanced thymoma. Radiother. Oncol. 19: 273-280. 1990. Verley. J.M. and Hollmann. K.H. Thymoma: a comparative study of clinical stages, histologic features and survival in 200 cases. Cancer 55: 1074-1086, 1985. Weissberg. D.. Goldberg. M. and Pearson. F.G. Thymoma. Ann. Thorac. Surg. 16: 141-147. 1973. Wilkins, E.W., Edmunds. L.H. and Castleman. B. Cases of thymoma at the Massachusetts General Hospital. J. Thorac. Cardiovasc. Surg. 52: 322-330. 1966. Wilkins, E.W., Grille, H.C.. Scannell, J.G.. Moncure, A.C. and Mathisen, D.J. Role of staging in prognosis and management of thymoma. Ann. Thorac. Surg. 51: 888-892, 1991.