Klaus Junker, MD; Kathrin Langner, MD; Folker Klinke, MD; Ulrich Bosse, MD; and Michael Thomas, MD

Grading of Tumor Regression in Non-small Cell Lung Cancer* Morphology and Prognosis Klaus Junker, MD; Kathrin Langner, MD; Folker Klinke, MD; Ulrich Bosse, MD; and Michael Thomas, MD Objective: Different types of multimodality therapy, including chemoradiotherapy and surgery, increasingly are being used for the treatment of patients with locally advanced non-small cell lung cancer (NSCLC; stages IIIA and IIIB). In this context, the applicability of a morphologic regression grading and its prognostic value were investigated. Patients and methods: In a multicenter phase II trial, 54 patients with locally advanced NSCLC received neoadjuvant bimodality treatment (ie, two cycles of ifosfamide, carboplatin, and etoposide, followed by twice-daily radiation up to 45 Gy with simultaneous administration of carboplatin and vindesine). Forty patients underwent resections. Using the corresponding resection specimens of the primary and regional lymph nodes, the following regression grading was established: grade I, no regression or only spontaneous tumor regression; grade II, morphologic evidence of therapy-induced tumor regression with at least 10% (grade IIa) or < 10% (grade IIb) vital tumor tissue; and grade III, complete tumor regression with no evidence of vital tumor tissue. Regression grading then was correlated with the survival time. Results: Three tumors were classified as regression grade I, 10 were classified as regression grade IIa, 20 were classified as regression grade IIb, and 7 were classified as regression grade III. Patients with tumors of regression grades IIb or III showed significantly longer survival times than those with tumors of regression grades I or IIa (median survival time, 36 vs 14 months, respectively; 3-year survival rate, 52% vs 9%, respectively; p 0.02). These survival times were also compared for patients who had undergone complete resection (median survival time, not reached vs 23 months, respectively; 3-year survival rate, 56% vs 11%, respectively; p 0.03). The presurgical clinical response after patients had received neoadjuvant multimodality therapy had no predictive value in assessing the extent of therapy-induced tumor regression in the resection specimen. Conclusions: After neoadjuvant therapy of patients with NSCLC, the proposed tumor regression grading was of predictive value for long-term survival. Beyond the achievement of complete tumor resection (R0), a therapy-induced tumor regression of < 10% of vital tumor tissue is pivotal for superior long-term outcomes. (CHEST 2001; 120:1584 1591) Key words: mediastinal lymph node downstaging; neoadjuvant therapy; non-small cell lung cancer; prognostic factors; regression grading Abbreviations: NSCLC non-small cell lung cancer; SWOG Southwest Oncology Group Conventional therapy of stage III non-small cell lung cancer (NSCLC) by surgery or radiation alone results in survival rates of only 5 to 13%. 1 These poor survival rates can be significantly improved by multimodality treatment using neoadjuvant chemotherapy or chemoradiotherapy followed by surgery. 2 Previous studies 2 4 on neoadjuvant therapy have demonstrated that the degree of tumor regression in resection specimens of the primary tumor and lymph nodes is suggestive for long-term survival. To determine whether reproducible pathologic-anatomic findings of therapy-induced tumor regression could be demonstrated, resection samples of locally advanced NSCLC after neoadjuvant mul- *From the Institute of Pathology (Drs. Junker and Langner), Bergmannsheil University Hospital, Bochum, Germany; Department of Thoracic Surgery (Prof. Klinke), St. Raphael Hospital, Ostercappeln, Germany; Institute of Pathology (Dr. Bosse), Osnabrück, Germany; and Department of Hematology (Dr. Thomas), Oncology and Respiratory Medicine, University of Münster, Münster, Germany. This research was supported by the North Rhine Westfalian Cancer Society, Düsseldorf, Germany. Manuscript received September 18, 2000; revision accepted June 14, 2001. Corresponding author: Klaus Junker, MD, Institute of Pathology, Bergmannsheil University Hospital, Bürkle-de-la-Camp-Platz 1, D- 44789 Bochum, Germany; e-mail: klaus.junker@ruhr-uni-bochum.de 1584 Clinical Investigations

timodality treatment were analyzed. For the first time in this treatment approach, a previously published 5 regression-grading scheme was applied, and its impact on survival was assessed. Materials and Methods Eligibility Criteria The study was approved by the ethical committee of the University of Münster, Germany. From April 1992 to September 1995, patients with NSCLC stage IIIA (tumor stage T1 3N2M0) with histologically confirmed N2 status or stage IIIB (tumor stage T4N1 3M0/T1 4N3M0) were enrolled in the study. Patients with involvement of the supraclavicular lymph nodes or who had positive test results for pleural effusion were not eligible. Further requirements were a favorable medical condition (Eastern Cooperative Oncology Group grade, 0 or 1), age range of 18 to 69 years, sufficient bone marrow reserve (leukocyte level, 4,000 cells/ L; thrombocyte level, 100,000 cells/ L), and adequate liver and kidney function (bilirubin level, 1.5 mg/dl; creatinine 1.5 mg/dl; creatinine clearance, 30 ml/min). The pretreatment evaluation included a laboratory test parameter profile, bronchoscopy, and image diagnosis, with chest radiograph, CT scan of the thorax, abdomen, and brain, and a bone scan. Mediastinoscopy was performed to assess the mediastinal lymph node status. Patients also were enrolled after an assessment of the mediastinal lymph nodes by means of an exploratory thoracotomy. Treatment Protocol Treatment started with two cycles of chemotherapy (the second cycle started on day 22) with carboplatin (300 mg/m 2 ; day 1), ifosfamide (1,500 mg/m 2 ; days 1, 3, and 5), and etoposide (100 mg/m 2 ; days 1, 3, and 5). The subcutaneous administration of granulocyte colony-stimulating factor (300 g/d) was started on day 6 and was continued until the leukocyte level exceeded 4,000 cells/ L. Three weeks after the start of the second cycle of chemotherapy, concurrent radiochemotherapy was commenced. Two 1.5-Gy fractions per day, with an intertreatment interval of at least 6 h, were administered 5 days a week to a total dose of 45 Gy (days 43 to 61). Carboplatin (100 mg/m 2 ) and vindesine (3 mg absolute) were administered on days 43, 50, and 57. Seven weeks, on average, after the completion of combined chemoradiotherapy, patients who remained free of distant metastases were eligible for thoracotomy. Surgery was performed with the objective of achieving complete resection of the tumor (ie, resection margins microscopically free of tumor cells with complete resection) and extensive mediastinal lymph node sampling. Patients with incomplete resections or inoperable tumors received further radiotherapy with conventional fractionation (16 Gy; 2 Gy 5 times per week) [Fig 1]. Response Assessment and Follow-up A preoperative assessment was performed 4 to 5 weeks after the completion of the combined chemoradiation and comprised a laboratory test parameter profile, bronchoscopy, chest radiograph, bone scan, and CT scans of the thorax, abdomen, and brain (see Fig 1 for restaging). Clinical response was assessed according to the guidelines of the Southwest Oncology Group (SWOG). 6 Follow-up examinations (ie, clinical test parameter profile, chest radiograph, and abdominal ultrasound) were undertaken every 3 months during the first 2 years after final staging and then every 6 months. CT scans of the thorax and brain were performed at 6, 12, 18, and 24 months after final staging. Regression Grading Figure 1. Scheme of treatment protocol. The surgical specimens were morphologically analyzed regarding therapy-induced changes of the tumor tissue. Initially, the formalin-fixed resection samples were inspected macroscopically. From the primary lesion, regions with likely vital tumor tissue or former, now regressively altered, tumor tissue as well as all resected mediastinal lymph nodes were embedded in paraffin. Depending on the size of the tumor, up to 58 paraffin blocks were processed histologically. For further evaluation, histologic slides of the macroscopically tumor-free surrounding lung parenchyma also were prepared. Hematoxylin-eosin and van Gieson stains were available for analysis of the specimens. In order to determine the degree of tumor regression, the type and extent of the vital tumor tissue, and the degree of tumor necrosis as well as reactive alterations with foam cell reaction and fibrosis or scar formation were taken into account. These findings were classified according to the following regression grading: CHEST / 120 / 5/ NOVEMBER, 2001 1585

Grade I: no tumor regression or only spontaneous tumor regression in the sections of the primary lesion and mediastinal lymph nodes; Grade II: morphologic evidence of therapy-induced tumor regression with at least 10% residual tumor cells in the sections of the primary lesion and/or mediastinal lymph nodes presenting more than focal microscopic disease (grade IIa) or 10% residual tumor cells in the sections of the primary lesion and/or mediastinal lymph nodes presenting focal microscopic disease (grade IIb); and Grade III: complete tumor regression with no evidence of vital tumor tissue in the sections of the primary lesion and mediastinal lymph nodes. Downstaging By means of the surgical specimens that were obtained, postoperative tumor stage and pathologic lymph node stage were established and were compared to the corresponding pretreatment data. Statistical Analysis The tumors of regression grades I and IIa and those of grades IIb and III were grouped together. After calculating the survival curves according to Kaplan and Meier, 7 the median survival time of both patient groups was compared (log-rank test). 8 The effect of downstaging on the overall survival time was checked using the same methods. Correlations between regression grade and histologic tumor type, clinical response status, or downstaging parameters were analyzed (Fisher s Exact Test). The relative importance of factors with significant influence on survival in a univariate analysis was estimated by multivariate analysis using the Cox proportional hazards regression model. 9 Because of the limited population, multivariate analysis was carried out in an explorative approach. Statistical significance was assumed at p 0.05. Results From April 1992 to September 1995, 54 patients (5 women and 49 men) with a median age of 57 years (age range, 37 to 69 years) were enrolled in the study. Twenty-five patients had stage IIIA disease, and 29 patients had stage IIIB disease. Histology identified 36 tumors as squamous cell carcinomas and 18 tumors as adenocarcinomas (Table 1). In 40 patients in the study population, surgery was performed after the completion of induction therapy. Of the remaining 14 patients, 3 were technically inoperable and 2 were functionally inoperable after radiochemotherapy. Three patients developed distant metastases during therapy. Another four patients refused surgery at that point. One patient died of pneumonitis, and one died due to tumor bleeding prior to surgery. Table 1 Patient Characteristics and Response to Treatment* Variables IIIA (n 25) Tumor Stage IIIB (n 29) III (n 54) Age, yr Median 59 57 57 Range 37 69 37 68 37 69 Sex Female 2 3 5 Male 23 26 49 Histology Squamous cell carcinoma 19 17 36 Adenocarcinoma 6 12 18 Clinical response CR 4 1 5 PR 15 17 32 NC 4 9 13 PD 2 2 4 *CR complete response; PR partial response; NC no change; PD progressive disease. Regression Grading Of 40 resection specimens available for further histologic assessment, 26 were squamous cell carcinomas and 14 were adenocarcinomas. According to the pretherapeutic staging, 20 tumors were classified as stage IIIA and 20 were classified as stage IIIB (Table 2). Four of these patients were women, 36 were men, and the median age was 57 years. In 37 of 40 resected tumors (92.5%), morphologic changes indicating therapy-induced tumor regression were seen after surgery. Among these, seven resection specimens (17.5%) showed complete tumor regression with no evidence of vital tumor tissue (ie, regression grade III). Circumscribed tumor residues with 10% vital tumor tissue were demonstrated in 20 samples (50%) [ie, regression grade IIb]. Ten resection specimens (25%) showed at least 10% vital tumor tissue (ie, regression grade IIa). In three cases (7.5%), no evidence of therapy-induced Table 2 Histomorphologic Grade of Tumor Regression in the Resection Specimens of the Study Patients* Variables Regression Grade I IIa IIb III Total Squamous cell carcinoma 2 5 16 3 26 Adenocarcinoma 1 5 4 4 14 Stage IIIA 1 5 11 3 20 Stage IIIB 2 5 9 4 20 CR 1 3 4 PR 1 8 13 6 28 NC 1 2 4 1 8 Total 3 10 20 7 40 *Tumor regressions are listed according to histologic type, clinical stage, and clinical response (n 40). See Table 1 for abbreviations not used in the text. 1586 Clinical Investigations

tumor regression (ie, regression grade I) was found (Table 2). In two tumors classified as regression grade IIb, small tumor residues were seen only in the mediastinal lymph nodes, while the former primary tumor revealed no remaining vital tumor tissue. Grouped together, 27 tumors (67.5%) showed regression grades IIb or III and 13 tumors (32.5%) showed regression grades I or IIa. This grouping was necessary for statistical evaluation because of groups too small for separate analysis. While 71.4% of squamous cell carcinomas were classified as having regression grades IIb or III, those regression grades were seen in only 58.3% of adenocarcinomas. The remaining tumors (28.6% of squamous cell carcinomas and 41.7% of adenocarcinomas) were of regression grades I or IIa. However, no significant correlation between histologic type and regression grade was demonstrated (p 0.28 [Fisher s Exact Test]). Figure 2. Target-like focus with centrally located, therapyinduced necrosis, surrounding foam cell rim, transition into vascular granulation tissue, and peripheral scar formation. Morphology of Therapy-Induced Tumor Regression In 24 patients with tumors of regression grades IIa to III, different sized target-like foci with central necrosis, narrow foam cell rim, vascular granulation tissue, and marked peripheral scarring were demonstrated in the tumor region. In addition, foam cell nests without central necrosis were present in four resection specimens. Toward the periphery, the foam cell rims and nests showed transition into vascular granulation tissue and adjoining cicatrization, sometimes forming large scarred areas and showing a focally accentuated increase or condensation of elastic fibers (Fig 2). In eight specimens (six adenocarcinomas and two squamous cell carcinomas) of regression grades IIa to III, scarred fibrosis of the former tumor tissue was the predominant morphologic alteration. Comparatively large tumor areas were present in the three samples that were classified as regression grade I. The areas of tumor necrosis seen in these cases did not exceed the extent of spontaneous tumor regression. In tumors that were classified as regression grade IIa after neoadjuvant therapy, signs of spontaneous and therapy-induced tumor regression were found simultaneously. Correlation of Morphologic Findings to Time- Related Course of Neoadjuvant Therapy For the 40 surgically treated patients, the presurgical therapy period ranged from 97 to 172 days (ie, from the first day of the first cycle of chemotherapy until the day of surgery), with an average of 123 days. The time intervals between the last day of radiochemotherapy and the date of surgery varied from 21 to 96 days, with 49 days being the average. There was no statistically significant correlation between the variation of the overall therapy periods or the variation of the intervals between the last day of radiochemotherapy and the date of surgery with the different grades of tumor regression (regression grades I/IIa vs IIb/III [Mann-Whitney U test]). The two patients with the longest overall therapy period (172 and 168 days) also showed the longest time interval between radiochemotherapy and surgery (90 and 96 days, respectively). The corresponding resection samples were classified as regression grade IIa with 10% vital tumor tissue. The morphologic changes seen here could not be separated from those observed in patients with markedly shorter therapy periods. Presurgical Clinical Response Based on the response criteria of the SWOG, 6 in this prognostically unfavorable group of 54 patients, efficacy in terms of response was seen with a response rate of 68.5% and a low 7.4% rate of progressive disease (Table 1). Of the group of 40 surgically treated patients, 28 showed partial responses, 4 showed complete responses, and 8 showed no change. Among resected tumors in these latter eight patients, five tumors revealed 10% vital tumor tissue (ie, regression grades IIb/III), whereas only three tumors with 10% vital tumor tissue were classified as regression grades I or IIa (Table 2). As an illustration, there was a 48-year-old female patient CHEST / 120 / 5/ NOVEMBER, 2001 1587

with progressive disease after chemotherapy and no change after radiochemotherapy. Macroscopically, the corresponding resection specimen included a large tumor that was 7 cm in diameter, which, on histologic examination, proved to be subtotally necrotic with 10% vital tumor tissue (regression grade IIb). On the other hand, there was one tumor with complete clinical response and one tumor with partial clinical response but that was missing evidence of therapy-induced tumor regression in the corresponding resection specimens (Table 2). In both cases, the pretherapy tumor stage was confirmed postsurgically. Both tumors revealed mediastinal infiltration and persistent stage N2 disease. Initially, patients with these tumors presented with a hilar mass and enlarged mediastinal lymph nodes. After preoperative induction, the central mass was reduced in one patient to a paraesophageal mediastinal infiltration, fulfilling the criteria for partial response, and in the other patient it was reduced to a slight paramediastinal thickening, fulfilling the criteria for complete response. Retrospectively, the clinical course of these patients was highly suspicious of a surrounding inflammatory consolidation that dissolved with treatment. Altogether, the presurgical response status after neoadjuvant multimodal therapy was of no predictive value for assessing the extent of therapy-induced tumor regression in the corresponding resection specimens (Fisher s Exact Test; p 1.0). State of Resection After the end of chemoradiotherapy a total of 40 patients underwent tumor resection with extensive mediastinal lymph node sampling. In 34 cases, complete resection (R0 resection) was achieved. Four patients had microscopic tumor involvement of the resection margin (R1 resection), and in two patients macroscopic tumor residues were left intraoperatively (R2 resection). Survival and Prognostic Impact of Regression Grading With a median follow-up period of 44 months, the median survival time for all enrolled patients was 20.4 months, and the 3-year survival rate amounted to 30%. For patients who underwent resection (n 40), the median survival time was 23.0 months, and the 3-year survival rate amounted to 36%. Neither patients age (ie, 60 years vs 60 years), histologic tumor type (squamous cell carcinoma vs adenocarcinoma), or disease stage (stage IIIA vs stage IIIB) were revealed as significant predictors for survival on univariate analysis. In patients who had undergone resection (n 40), the morphologic regression grading was applied as described. Patients classified as having regression grades IIb and III (n 27) showed a significantly longer survival time compared to those who were classified as having regression grades I and IIa (n 13) (median survival time, 36 vs 14 months, respectively; 3-year survival rate, 52% vs 9%, respectively; p 0.02 [log-rank test]) (Fig 3). Moreover, in patients who had undergone complete tumor resection (n 34), survival periods could be shown to be significantly longer than in those with incomplete resection or no surgery at all (n 20). Even in patients who had undergone complete resections, the grade of tumor regression (ie, regression grades IIb/III vs I/IIa) remained a significant predictor for survival (median survival time, not reached vs 23 months, respectively; 3-year survival rate, 56% vs 11%, respectively; p 0.03 [log-rank test]). Downstaging In comparison to the initial clinical staging, 28 of 40 surgically treated patients (70%) showed a lower tumor stage in the resection specimens (Table 3). In 29 patients (72.5%), a lower N-stage tumor was seen, while 11 tumors (27.5%) maintained their initial lymph node status (Table 3). To assess the impact of mediastinal downstaging on survival, the survival periods of patients with histologically documented postsurgical stage N0 or N1 lymph node status (n 26) were compared to those with persistent mediastinal lymph node involvement (N2 disease; n 14). Here, different median overall survival periods of 29 and 23 months, respectively, could be established without reaching statistical significance (p 0.27, log-rank test). On Figure 3. Survival rate from the time of diagnosis in patients with tumor resection (n 40) according to postoperative regression grading (ie, regression grades I/IIa vs IIb/III; median survival time, 14 vs 36 months, respectively; 3-year survival rate, 9% vs 52%, respectively; p 0.02). 1588 Clinical Investigations

Table 3 Comparison of Pretherapy and Postsurgical Tumor Stage and Lymph Node Status* Pre 3 Post No. % Tumor stage IIIB 3 IIIB 2 5 IIIB 3 IIIA 3 7.5 IIIB 3 II 6 15 IIIB 3 I 5 12.5 IIIB 3 0 4 10 IIIA 3 IIIA 10 25 IIIA 3 II 2 5 IIIA 3 I 5 12.5 IIIA 3 0 3 7.5 Total 40 100 Lymph node status N3 3 N3 N3 3 N2 4 10 N3 3 N1 3 7.5 N3 3 N0 4 10 N2 3 N2 10 25 N2 3 N1 5 12.5 N2 3 N0 11 27.5 N1 3 N1 N1 3 N0 2 5 N0 3 N0 1 2.5 Total 40 100 *Pre presurgery; Post postsurgery. the other hand, postsurgical lymph node status and the extent of therapy-induced tumor regression in the corresponding resection specimens showed a significant correlation (p 0.02, Fisher s Exact Test). Of 26 patients with postsurgical stage N0 or N1 lymph node status, 21 resection specimens (primary tumors and lymph nodes) revealed 10% vital tumor tissue (ie, regression grades IIb/III). In univariate analyses, the clinical response to preoperative induction therapy, resection status, and the extent of therapy-induced tumor regression were found to be significant predictors for survival. For patients who had undergone tumor resections (n 40), a step-by-step multivariate analysis according to the Cox proportional hazards regression model was applied in an explorative approach. Beyond the grade of tumor regression (regression grades I/IIa vs IIb/III; p 0.007), the resection status (R0 vs non- R0; p 0.009) was revealed as an independent predictor for survival. This could not be established for clinical response (complete response/partial response vs no change/progressive disease; p 0.1). Discussion In a phase II study, 54 patients with locally advanced NSCLC (stages IIIA and IIIB) were treated with neoadjuvant chemotherapy followed by radiochemotherapy. In 40 of these patients, surgery was performed. 10 Comparing the population analyzed here with 50 patients with previously untreated NSCLCs, we were able to show that spontaneous and therapy-induced tumor regression can be distinguished with high certainty. 5 Several morphologic changes that were due to neoadjuvant chemoradiotherapy were demonstrated. However, it has to be pointed out that the single changes are merely nonspecific but, in their entirety, enable us to draw reliable conclusions regarding the response to presurgical therapy. No correlation between the time-related course of neoadjuvant therapy and the occurrence of morphologic changes indicating therapy-induced tumor regression or the corresponding regression grade was shown. Because of the great variability in the time intervals between the last day of radiochemotherapy and the date of surgery (21 to 96 days), one might assume the established regression grade to be not only an effect of therapy-induced tumor regression but also of the differences in tumor cell progression after the end of radiochemotherapy. Nonetheless, there was no correlation between the variation of the preoperative intertreatment interval (ie, the period from the end of radiochemotherapy to surgery) with the different grades of tumor regression. Moreover, according to unpublished data, differences in tumor cell proliferation determined by the Ki67 antigen-labeling index or the extent of apoptosis (terminal deoxynucleotidyl transferase-mediated dutp nick end labeling technique) did not correlate with different regression grades (regression grade I/IIa vs IIb/III, Mann-Whitney U test). Statistical evaluation showed that patients with 10% vital tumor tissue (ie, regression grades IIb/ III) had a statistically significant longer median survival time (36 months) than patients with regression grades I or IIa (14 months) (p 0.02, log-rank test). In an explorative approach, for those patients with tumor resection the extent of tumor regression (regression grades I/IIa vs IIb/III) proved to be a statistically independent prognostic factor. Bromley and Szur 11 reported on 66 cases of resected NSCLC after presurgical radiotherapy. Twenty-nine of the corresponding resection specimens did not show any vital tumor tissue. This finding was not correlated to higher survival rates in the investigated population. Also, Shields et al 12 were not able to show survival advantages for patients who had undergone presurgical radiotherapy of lung cancer despite histologically complete tumor regression. However, after the application of neoadjuvant chemotherapy, alone or in combination with radiotherapy, a positive correlation between histologically confirmed complete tumor regression and a favor- CHEST / 120 / 5/ NOVEMBER, 2001 1589

able prognosis has been discussed. 13 19 Moreover, several trials demonstrated a significant survival advantage according to the extent of histomorphologic response after chemoradiation. 3,4,10,20 A high rate of pathologic complete response of 39% was reported by Eberhardt et al 21 in the resection specimens of locally advanced NSCLC after preoperative chemoradiotherapy. In contrast to the trials mentioned, after undergoing complete resection, no differences in long-term survival times were found between patients who had a pathologic complete response and those with persistent viable tumors. However, Eberhardt and colleagues 21 did not perform a standardized histologic examination of the entire tumor area. Basically, the determination of complete or predominant tumor regression immediately after the completion of neoadjuvant therapy may offer information on the potential benefits of a particular type of therapy. So, this parameter may be included as an intermediate end point in future neoadjuvant therapy trials involving patients who have stage III NSCLC. 3,20 However, complete step-by-step processing of the primary tumor area in the resection specimens and of the resected mediastinal lymph nodes is mandatory. A favorable or unfavorable value of the therapy concept possibly may be determined immediately after completion of the therapy and before reaching a certain follow-up period. Thus, following studies may be started earlier. A standardized classification of tumor regression may also assist in creating an unambiguous description and, subsequently, a better comparability of this parameter in different studies. A discordance between clinical response to neoadjuvant multimodality therapy and the extent of histomorphologically determined tumor regression has been discussed by several authors. 22,23 Additionally, in the present study, no correlation was found between the clinical assessment of the response to neoadjuvant therapy after radiochemotherapy and the grade of tumor regression (p 1.0, Fisher s Exact Test). More than half of the tumors showing no change according to the remission criteria of the SWOG revealed morphologically marked therapyinduced tumor regression with 10% vital tumor tissue. The predominant cause for this discrepancy is the fact that even in CT scans vital tumor cannot be differentiated from already necrotic tumor tissue or scar formations. After neoadjuvant therapy, tumors of several centimeters may show marked or even complete tumor regression, so that a generous indication for surgery is justified in cases of no change after therapy. The currently used imaging techniques do not allow the reliable determination of therapy success during neoadjuvant therapy. The question of mediastinal downstaging is of great clinical interest. In contrast to the regression grading system applied here, it could potentially be evaluated prior to surgery by means of mediastinoscopy. The impact of mediastinal downstaging was addressed with a detailed workup in the Massachusetts General Hospital trial by Choi et al, 3 who investigated a population of 42 patients with stage IIIA(N2) NSCLC who had undergone preoperative twice-daily radiation therapy and concurrent chemotherapy. In good correlation with our results, they found a lymph node downstaging in 67% of the analyzed tumors (n 28). Corresponding to our own investigations, Choi et al 3 could not show a statistically significant difference in survival between postsurgical stages II (N1) and III (N2), despite different 5-year survival rates of 42% and 18%, respectively. Here, significantly different survival periods were only established comparing postthoracotomy stages 0 and I (N0) vs III (N2) [p 0.04]. Altogether, the grading of therapy-induced tumor regression, considering both therapy-induced effects in the primary tumor and in lymph node metastases, seems to be a more precise method for predicting the outcome of the disease than the exclusive assessment of mediastinal downstaging. In conclusion, it has been shown that in resection specimens of stage III NSCLC after neoadjuvant therapy the presence of marked tumor regression with 10% vital tumor tissue (ie, regression grades IIb/III) is predictive for superior survival times of these patients. The applied regression grading system is an independent prognostic factor in locally advanced lung cancer. The decisive parameters of success in neoadjuvant therapy for patients with locally advanced NSCLC are the achievement of complete tumor resection (R0 status) and total, or at least subtotal, therapy-induced tumor regression. ACKNOWLEDGMENTS: The authors thank Professor W. Böcker (Münster, Germany), Professor K.-F. Bürrig (Hildesheim, Germany), Dr. W.-P. Kunze (Hemer, Germany), and Professor W. Lang (Hannover, Germany) for their kind support in contributing biopsy and resection specimens to this study, and Dr. A. Heinecke (Münster, Germany) for his support in statistical analysis. This article is dedicated to Professor Dr. K.-M. Müller (Bochum, Germany) on the occasion of his 60th birthday. References 1 Mountain CF. Revisions in the international system for staging lung cancer. Chest 1997; 111:1710 1717 2 Johnson DH, Turrisi AT, Pass HI. Combined modality treatment for locally advanced non-small cell lung cancer. In: Pass HI, Mitchell JB, Johnson DH, et al, eds. Lung cancer: principles and practice. New York, NY: Lippincott, 1996; 863 873 3 Choi NC, Carey RW, Daly W, et al. Potential impact on survival of improved tumor downstaging and resection rate by preoperative twice-daily radiation and concurrent chemother- 1590 Clinical Investigations

apy in stage IIIA non-small cell lung cancer. J Clin Oncol 1997; 15:712 722 4 Albain KS, Rusch VW, Crowley JJ, et al. Concurrent cisplatin/ etoposide plus chest radiotherapy followed by surgery for stages IIIA (N2) and IIIB non-small cell lung cancer: mature results of Southwest Oncology Group phase II study 8805. J Clin Oncol 1995; 13:1880 1892 5 Junker K, Thomas M, Schulmann K, et al. Tumour regression in non-small cell lung cancer following neoadjuvant therapy: histological assessment. J Cancer Res Clin Oncol 1997; 123:469 477 6 Green S, Weiss GR. Southwest Oncology Group standard response criteria, endpoint definitions and toxicity criteria. West New Drugs 1992; 10:243 253 7 Kaplan EL, Meier P. Non-parametrics. estimation from incomplete observations. J Am Stat Assoc 1958; 53:457 481 8 Peto R, Pike MC, Armitage P, et al. Design and analysis of randomized clinical trials requiring prolonged observation of each patient: II. Analysis and examples. Br J Cancer 1977; 35:1 39 9 Cox DR. Regression models and life tables (with discussion). J R Stat Soc B 1972; 34:187 220 10 Thomas M, Rübe C, Semik M, et al. Impact of preoperative bimodality induction including twice-daily radiation on tumor regression and survival in stage III non-small cell lung cancer. J Clin Oncol 1999; 17:1185 1193 11 Bromley LL, Szur L. Combined radiotherapy and resection for carcinoma of the bronchus. Lancet 1955:937 941 12 Shields TW, Higgins GA, Lawton R. Preoperative x-ray therapy as an adjuvant in the treatment of bronchogenic carcinoma. J Thorac Cardiovasc Surg 1970; 59:49 61 13 Burkes RL, Ginsberg RJ, Shepherd FA. Induction chemotherapy with mitomycin, vindesine, and cisplatin for stage III unresectable non-small cell lung cancer: results of the Toronto phase II trial. J Clin Oncol 1992; 10:580 586 14 Eagan R, Ruud C, Lee R. Pilot study of induction therapy with cyclophosphamide, doxorubicin, and cisplatin (CAP) and chest irradiation prior to thoracotomy in initially inoperable stage III M0 non-small cell lung cancer. Cancer Treat Rep 1987; 71:895 900 15 Faber LD, Kittle CF, Warren WT. Preoperative chemotherapy and irradiation for stage III non-small cell lung cancer. Ann Thorac Surg 1989; 47:669 677 16 Pisters KMW, Kris MG, Gralla RJ. Preoperative chemotherapy in stage IIIA non-small cell lung cancer: an analysis of a trial in patients with clinically apparent mediastinal node involvement. In: Salmon SE, ed. Adjuvant therapy of cancer. Philadelphia, PA: Saunders, 1990; 133 137 17 Skarin A, Jochelson M, Sheldon T. Neoadjuvant chemotherapy in marginally resectable stage III M0 non-small cell lung cancer: long-term follow-up in 41 patients. J Surg Oncol 1989; 40:266 274 18 Strauss G, Herndon JE, Sherman DD, et al. Neoadjuvant chemotherapy and radiotherapy followed by surgery in stage IIIA non-small cell carcinoma of the lung: report of cancer and leukemia group B phase II study. J Clin Oncol 1992; 10:1237 1244 19 Weiden P, Piantadosi S. Preoperative chemotherapy (cisplatin, and fluorouracil), and radiation in stage III. non-small cell lung cancer: a phase II study of the lung cancer study group. J Natl Cancer Inst 1991; 83:266 272 20 Pisters KMW, Kris MG, Gralla RJ, et al. Pathologic complete response in advanced non-small cell lung cancer following preoperative chemotherapy: implications for the design of future non-small cell lung cancer combined modality trials. J Clin Oncol 1993; 11:1757 1762 21 Eberhardt W, Wilke H, Stamatis G, et al. Preoperative chemotherapy followed by concurrent chemoradiation therapy based on hyperfractionated accelerated radiotherapy and definite surgery in locally advanced non-small cell lung cancer. J Clin Oncol 1998; 16:622 634 22 Edelmann MJ, Gandara DR, Roach M, et al. Multimodality therapy in stage III non-small cell lung cancer. Ann Thorac Surg 1996; 47:669 675 23 Rusch VW. Neoadjuvant chemotherapy for stage III lung cancer. Semin Thorac Cardiovasc Surg 1993; 5:258 267 CHEST / 120 / 5/ NOVEMBER, 2001 1591