The presence of lymph node metastasis is a crucial

Micrometastasis to Lymph Nodes in Stage I Left Lung Cancer Patients Ryoji Kawano, MD, Enjo Hata, MD, Shingo Ikeda, MD, and Hirozo Sakaguchi, MD Surgical Department of Respiratory Center, Mitsui Memorial Hospital, Tokyo, Japan Background. To evaluate the frequency and clinicopathological characteristics of lymph node micrometastasis in left lung cancer patients diagnosed to be stage IA and IB based on routine histopathologic examinations, we examined the lymph nodes in patients who had undergone an extended mediastinal lymphadenectomy, using immunohistochemical methods. Methods. Paraffin-embedded tissue sections from the lymph nodes in 49 patients with stage I left lung cancers were studied. We used AE1/AE3 as the anticytokeratin and Ber-EP4 as the antiepithelial cell antibodies when performing immunohistochemical staining. Results. We identified micrometastasis of the lymph nodes in 13 (26.5%) of 49 patients with stage I left lung cancer. N0 disease was reclassified as N1 disease in 5 cases, N2 disease in 6 cases, and N3 disease in 2 cases. The location of the micrometastatic lymph nodes proved to be wide regions including the contralateral and highest mediastinal nodes, and 6 (46.2%) out of the 13 patients with micrometastasis were thus presumed not to be completely eliminated by a standard lymphadenectomy through an ipsilateral thoracotomy. The five year survival rate of patients with reclassified N1 to N3 disease was 74%, and the presence of micrometastasis was found to have no significant effect on the outcomes. Conclusions. The micrometastatic involvement of the lymph nodes was both more frequent and extensive than expected even in stage I left lung cancer. These results suggest that an extended mediastinal lymphadenectomy may therefore be required for the locoregional control of stage I left lung cancer patients. (Ann Thorac Surg 2002;73:1558 62) 2002 by The Society of Thoracic Surgeons The presence of lymph node metastasis is a crucial prognostic factor for primary lung cancer. A microscopic evaluation of dissected lymph nodes allows for accurate staging, thus leading to the selection of appropriate additional therapies such as systemic chemotherapy or irradiation. However, it has been recognized that micrometastatic spread is not easy to identify by routine histopathologic examinations alone. Therefore, the incidence of nodal metastasis is currently underestimated using the present procedures; in other words, inappropriate staging leaves the biological characteristics of primary lung cancer unclear, and an underestimation of the disease stage may result in inappropriate therapy selection [1 6]. Recent advances in sensitive immunohistochemical techniques and specific monoclonal antibodies now allow detection of small clusters or even single tumor cells in the lymph nodes that normally appear to be negative on routine histopathological examinations. Most previous investigations have clearly shown that lung cancer patients with micrometastasis of the lymph nodes have a poorer prognosis than patients without such micrometastasis. As a result, the relatively high incidence of tumor recurrence even in stage I lung cancer patients is thus suggested to be correlated with the existence of Accepted for publication Dec 21, 2001. Address reprint requests to Dr Kawano, Surgical Department of Respiratory Center, Mitsui Memorial Hospital, Kandaizumi-cho 1, Chiyoda-ku, Tokyo, Japan; e-mail: 2ryo@msc.biglobe.ne.jp. occult micrometastasis in the lymph nodes [1 5]. A better understanding of the actual circumstances of micrometastasis in the lymph nodes is therefore important in recognizing the appropriate region of a mediastinal lymphadenectomy; and it may also help to clarify whether patients diagnosed to be pn0 by routine histopathologic examinations actually need to undergo lymphadenectomy. However, previously reported cases of micrometastasis include cases obtained by examining only the lymph nodes collected by a mediastinal lymphadenectomy (nodal dissection 2: ND2) through a standard thoracotomy. Based on this viewpoint, we report here the frequency and location of micrometastasis in the lymph nodes of left lung cancer patients with pt1n0m0 to pt2n0m0 disease who underwent an extended mediastinal lymphadenectomy. We have also tried to identify the optimal area of the lymph node dissection to obtain locoregional control in left lung cancer patients. Material and Methods We collected the removed lymph nodes and tumor samples from 49 left lung cancer patients diagnosed with T1N0M0 (n 26) and T2N0M0 (n 23) pathology who had been treated by either a lobectomy or a pneumonectomy in combination with an extended mediastinal lymphadenectomy between 1986 and 1998 at the Surgical Department of the Respiratory Center, Mitsui Memorial Hospital. Of 49 patients, 45 (91.8%) were preoperatively 2002 by The Society of Thoracic Surgeons 0003-4975/02/$22.00 Published by Elsevier Science Inc PII S0003-4975(02)03398-2

Ann Thorac Surg KAWANO ET AL 2002;73:1558 62 MICROMETASTASIS TO LYMPH NODES IN LUNG CANCER 1559 Table 1. Relationship Between Various Clinicopathologic Characteristics and Presence of Micrometastasis of Lymph Nodes Characteristic No. of Patients No. of Patients With Micrometastasis (%) Age, y (median 62.4) 65 25 5 (20.0) 65 24 8 (33.3) Sex Male 33 9 (27.2) Female 16 4 (25.0) T-factor T1 26 6 (23.0) T2 23 7 (30.4) Location of tumor Upper lobe 33 9 (27.2) Lower lobe 16 4 (25.0) Histology of tumor Adenocarcinoma 30 9 (30.0) Squamous cell carcinoma 19 4 (21.0) Differentiation of tumor Good 30 7 (23.3) Moderate and poor 19 6 (31.5) diagnosed as clinical stage N0. A complete resection was performed in all cases. The stage of the disease was based on the TNM classification of the International Union Against Cancer [7]. Clinicopathologic characteristics of the patients are shown in Table 1. The criteria for undergoing an extended operation for primary left lung cancer were age 75 years or less, no serious complications, and left lung cancer (except for cases of extensive chest wall invasion, regardless of clinical N factor). Briefly, the operative method was as follows. A median sternotomy was performed in the upper lobe of the left lung, and a median sternotomy and fifth intercostal anteroaxillary incision were then performed in the lower lobe of the lung. After the median sternotomy, the anterior mediastinal tissue, together with the thymus, is removed. The lymph nodes around the right and left recurrent laryngeal nerves directly under the thyroid gland, which is the upper limit of mediastinal dissection, followed by a series of lymph nodes on the bilateral sides along the trachea were then dissected. After performing a vertical pericardiotomy, an incision was made between the superior vena cava and the ascending aorta. Next, the leftward retraction of the ascending aorta was performed using a spatulate retractor, followed by a dissection of the lymph nodes in a downward direction as described above. The right side margin of the dissection was set at the right lower paratracheal nodes (4R). The mediastinal dissected tissue, including the left lower paratracheal nodes (4L) and subcarinal nodes, are passed under the aortic arch in sequence along the left main bronchus. On the side of the thoracic cavity, the lymph nodes along the vagus nerve, Botallo s ligament, and the paraaortic region are dissected. Finally, the mediastinal dissected tissue is subsequently transferred en block toward the hilum of the lung. The detailed theoretical background for this operative method is given elsewhere in detail [8 10]. The mean postoperative hospitalization period was 22.3 9.8 days. Surgically related death occurred in 1 patient, who died of respiratory insufficiency. Surgical complications included paresis of the recurrent laryngeal nerve in 6 patients (12.2%); however, all patients recovered within 1 to 3 months of follow-up. Other complications included pneumonia and supraventricular arrhythmia, although these were not severe. After the operation, the patients were reexamined once every 2 to 4 weeks for 2 years, and thereafter once every 2 or 3 months for 3 years. In addition, from the 6th year postoperatively, all patients were followed at biannual visits. The evaluations for local recurrence and distant metastasis were made by physical examinations, chest roentgenography, computed tomography (head, thoracic, and abdominal regions), and bone scintigraphy. The median follow-up duration of the patients was 66.6 months (range 1.5 to 164.9 months). Immunohistochemical Method Paraffin-embedded, formalin-fixed tissue sections 4- m thick were prepared for immunohistochemical staining in 10% buffered solution and then deparaffinized. We used monoclonal antibodies to broad spectrum cytokeratin (dilution 1:50, AE1/AE3, Dako Corporation, CA) and Ber-EP4 (dilution 1:50, Dakopatts, Denmark). The AE1/ AE3 antibody specifically recognizes the epithelial normal and tumor cells, whereas Ber-EP4 targets two epithelial cell surface glycopolypeptides [11]. An immunohistochemical analysis was performed using the EnVision kit (Dako Corporation, CA) based on the labeled polymer method. Briefly, this method was as follows: after blocking the endogenous peroxidase activity, background staining was removed. Monoclonal antibodies and a polymer reagent were then applied. We studied three sections taken from each lymph node block to detect any immunoreactive cells of cytokeratin and Ber-EP4 antibodies. The sections were cut from different levels of lymph nodes. Primary tumors were confirmed to demonstrate positive immunoreactivities against their antibodies. As a control, consecutive sections from cytokeratin and Ber-EP4 positive tumor and Ber-EP4 negative lymphoid tissue samples were stained under the same conditions. Statistical Analysis A proportion comparison was done using the two-tailed Fisher s exact test. The survival rate including all postoperative deaths (lung cancer related and unrelated) was calculated by the Kaplan-Meier method. The log-rank test was used to examine the statistical significance of any differences observed. A p value of less than 0.05 was considered to be statistically significant. The Cox proportional hazards model was applied for univariate and multivariate analyses. Statistical analysis was conducted using the Statview software package, version 5.0 (Abacus Concepts, Berkeley, CA).

1560 KAWANO ET AL Ann Thorac Surg MICROMETASTASIS TO LYMPH NODES IN LUNG CANCER 2002;73:1558 62 Fig 2. Micrometastatic sites in patients diagnosed to be stage I left lung cancer by routine histopathologic examination with hematoxylin and eosin staining. Fig 1. Immunoreactive tumor cells in lymph nodes against anticytokeratin antibody. Small cluster of tumor cells are seen in subcapsular sinus, indicated by arrow. (Original magnification 200.) Results Incidence and Location of Micrometastasis in Lymph Nodes In 49 cases, a total of 1,820 lymph nodes were examined. The incidence of micrometastasis in the dissected lymph nodes diagnosed to be N0 by routine histologic examinations of hematoxylin and eosin stained slides was 13 (26.5%) of 49 cases. Among the 26 cases with pt1n0m0 and 23 cases with pt2n0m0, positive cases of micrometastasis were detected in 6 cases (23.0%) and in 7 cases (30.4%), respectively. A positive finding of immunoreactivity against anticytokeratin antibody in the lymph nodes is shown in Figure 1. The lymph node stations showing micrometastases are depicted in Figure 2. Clinical and pathologic backgrounds of the patients with micrometastasis are shown in Table 1. There was no statistically significant correlation between the various clinicopathologic factors and the micrometastasis of lymph nodes. Stage IA and IB diseases were reclassified as stage IIA in 4 cases, IIB in 1 case, IIIA in 6 cases, and IIIB in 2 cases. From the results of a study on the site-based occurrence of micrometastasis in primary lesions, micrometastasis was detected in 5 (45.4%) of 11 cases in the lingular lobe of the lung, which occurred more frequently than in other segments. Skip micrometastasis was defined as metastatic foci in the mediastinum (N2 and N3) without metastases in either the hilar or lobar nodes (N1). As a result, skip micrometastasis was identified in 7 (53.8%) of 13 cases. Micrometastasis in 6 (46.2%) of 13 patients was found in the lymph nodes, which were impossible to dissect completely by means of a lymphadenectomy through a standard thoracotomy. The stations of these lymph nodes were the highest right mediastinal node (1R), 4R, highest left mediastinal node (1L), upper left paratracheal node (2L), prevascular node (3A), and 4L. In addition, of 13 patients with micrometastasis, 3 patients (23.0%) had a maximum tumor diameter of less than 2 cm. Immunohistochemical Results All primary tumors showed good immunoreactivity for AE1/AE3 and Ber-EP4 antibodies. The majority of the immunoreactive tumor cells existed in the subcapsular and paracortical sinuses of lymphoid tissue in single or small cluster conditions (Fig 1). Survival Curves of Patients With and Without Micrometastases in Lymph Nodes The overall 5- and 10-year survival rates in patients with stage I disease who underwent an extended mediastinal lymphadenectomy were 81.5% and 63.1%, respectively. The 5-year survival of patients with and without micrometastasis were 74.0% and 84.4%, respectively; as shown in Figure 3, there was no statistically significant difference between the two groups (p 0.541). Furthermore, no evidence of independent prognostic factors in micrometastatic patients was identified (data not shown), and the micrometastasis did not substantially affect the prognosis based on a multivariate analysis. Relationship Between Micrometastasis and Local Recurrence or Distant Metastasis Among the micrometastatic patients, distant metastasis occurred in 3 patients (23.0%) during the follow-up period. Of the 3 patients, 1 patient was restaged as stage IIIA and 2 were restaged as IIIB; all 3 patients developed multiple pulmonary and bone metastases and, as a result, eventually died. However, no local recurrence was found in any of the patients. On the other hand, there were 7 patients who demonstrated no micrometastases but nevertheless died during the follow-up period. Three patients had noncancerous diseases and 4 patients died due to distant metastasis

Ann Thorac Surg KAWANO ET AL 2002;73:1558 62 MICROMETASTASIS TO LYMPH NODES IN LUNG CANCER 1561 Fig 3. Survival curves in patients with and without micrometastasis of lymph nodes. (multiple bone and lung metastases) without locoregional recurrence. Comment Cytokeratin is a component of the intermediate filament that composes part of the cytoskeleton of epithelial cells [12]. Tumors derived from epithelial cells retain the characteristics of cytokeratin, and such tumors are therefore positive for cytokeratin with high specificity on immunostaining. Nevertheless, some cell types (eg, reticulum or mesothelial cells) in normal lymph nodes have been reported to give false-positive results [13 15]. However, differentiating between false-positive cells and tumor cells is possible using microscopic cytology or immunohistochemical characteristics against Ber-EP4 antibody [11]. In reports on micrometastasis, the rate of micrometastasis of lung cancer to lymph nodes ranged from 10.4% to 70.5% [1 6]. The range of the reported rates depended on tumor characteristics; therefore, study subjects should be matched for the clinical backgrounds of patients to be compared for the cytokeratin-positive rate. At any rate, it is both noteworthy and alarming that micrometastasis to the lymph nodes was detected at a higher frequency than expected among patients diagnosed to be N0 by microscopic evaluations with hematoxylin and eosin stain. Furthermore, our results suggest that micrometastasis can be present irrespective of the category of the tumor, even in small tumors. Based on this fact, a mediastinal lymphadenectomy should thus be performed even in patients diagnosed as clinical N0 to obtain a complete resection of lung cancer. Micrometastasis to the lymph nodes, especially single tumor cells, is considered to be a first step of lymphogenous metastasis of carcinoma and the earliest stage of growth of cancer cells in the lymph node. Micrometastases were extensively observed in the mediastinal N2 and N3 stations not only along the major lymphatic route but also along the minor route [8]. With reference to the site of metastasis reported by Hata and colleagues [8], lymph node metastases were mainly detected along the lymph drainage routes of the primary lesion. In the left lung, the lymphatic flow ascends over a wide area of the mediastinum, involving the routes along the bilateral paratrachea and the aortic arch. Especially, in the instance of lower lobe cancer, a considerable amount of its flow is cross-distributed along the right route. Furthermore, it is noteworthy that skip micrometastases occurred even in the micrometastatic stage. Accordingly, no lymph node was regarded as a sentinel node in cases of primary left lung cancer. On the other hand, it may be possible that a small number of metastatic tumor cells detected in the lymph nodes can be eliminated through the immune mechanisms in the body, and such tumor cells are only transiently observed in the lymph nodes. Although there have so far been no reports concerning the biological behavior of tumor cells in the local lymph nodes, the outcomes of patients with micrometastasis to the lymph nodes have been reported to be similar to those of disease-stage matched patients with overt metastasis by hematoxylin and eosin staining. Therefore, the presence of tumor cells in the lymph nodes represents established metastasis, although these cells showed the same biological behavior as those in overt lymph node metastasis. In contrast to this fact, the presence of micrometastasis to the lymph nodes had no significant effect on patient outcomes in the present study. We assume the reason why our results showed a more favorable outcome than those in previous reports is as follows. First, the extended mediastinal lymphadenectomy, which is performed in regions in which tumor cells cannot be eliminated by a lymphadenectomy through a standard thoracotomy, appears to contribute to favorable outcomes. In fact, this extended operation results in an improvement in the survival of left lung cancer patients with N2 to N3 disease. For instance, the 5-year survival rate in 74 patients with pn2 to pn3 disease who underwent this procedure was 41.3% [10]. From this viewpoint, an extended mediastinal lymphadenectomy may be essential for a curative operation so as to obtain sufficient local control in left lung cancer patients. However, the 3 deaths caused by distant metastasis among the patients with micrometastasis should also be taken into careful consideration regarding the local control of lung cancer. The second reason for our relatively favorable results is that our study only evaluated a limited number of patients with micrometastases. Further investigations in a larger group of patients are needed to compare the outcomes between them. In conclusion, the main findings of this study are that: (1) micrometastasis to lymph nodes in stage I left lung cancer patients was observed at a rate of 26.5%; and (2) micrometastasis to the lymph nodes was extensively distributed in the mediastinal nodal stations. Therefore, an extended mediastinal lymphadenectomy may be required for the complete resection of cancer cells in stage I left lung cancer patients.

1562 KAWANO ET AL Ann Thorac Surg MICROMETASTASIS TO LYMPH NODES IN LUNG CANCER 2002;73:1558 62 References 1. Chen ZL, Perez S, Holmes EC, et al. Frequency and distribution of occult micrometastases in lymph nodes of patients with non-small cell lung carcinoma. J Natl Cancer Inst 1993;85:493 98. 2. Passlick B, Izbicki JR, Kubuschok B, et al. Immunohistochemical assessment of individual tumor cells in lymph nodes of patients with non-small-cell lung cancer. J Clin Oncol 1994;12:1827 32. 3. Izbicki JR, Passlick B, Hosch SB, et al. Mode of spread in the early phase of lymphatic metastasis in non-small-cell lung cancer: significance of nodal micrometastasis. J Thorac Cardiovasc Surg 1996;112:623 30. 4. Passlick B, Izbicki JR, Kubuschok B, et al. Detection of disseminated lung cancer cells in lymph nodes: impact on staging and prognosis. Ann Thorac Surg 1996;61:177 83. 5. Maruyama R, Sugio K, Mitsudomi T, et al. Relationship between early recurrence and micrometastases in the lymph nodes of patients with stage I non-small-cell lung cancer. J Thorac Cardiovasc Surg 1997;114:535 43. 6. Nicholson AG, Graham ANJ, Pezzella F, et al. Does the use of immunohistochemistry to identify micrometastases provide useful information in the staging of node-negative non-small cell lung carcinomas? Lung Cancer 1997;18:231 40. 7. Sobin LH, Wittekind CM, eds. International Union Against Cancer (UICC): TNM classification of malignant tumours, 5th ed. New York: John Wiley, 1997:91 7. 8. Hata E, Hayakawa K, Miyamoto H, et al. Rationale for extended lymphadenectomy for lung cancer. Theor Surg 1990;5:19 25. 9. Hata E, Hayakawa K, Miyamoto H, et al. The incidence and prognosis of the contralateral mediastinal node involvement of the left lung cancer patients who underwent bilateral mediastinal nodal dissection and pulmonary resection through a median sternotomy [Abstract]. Lung Cancer 1988;4:A87. 10. Hata E, Miyamoto H, Sakao Y. Investigation into mediastinal lymph node metastasis of lung cancer and rationale for decision of the extent of mediastinal dissection [Abstract]. J Jpn Surg Soc 1997;98:8 15. 11. Latza U, Niedobitek G, Schwarting R, et al. Ber-Ep4: new monoclonal antibody which distinguishes epithelia from mesothelia. J Clin Pathol 1990;43:213 9. 12. Moll R, Franke WW, Schiler DL, et al. The catalog of human cytokeratins: patterns of expression in normal epithelia tumors and cultured cells. Cell 1982;31:11 24. 13. Traweek ST, Liu J, Battifora H. Keratin gene expression in non-epithelial tissues: detection with polymerase chain reaction. Am J Pathol 1993;142:1111 8. 14. Rutty GN, Lauder I. Mesothelial cell inclusions within mediastinal lymph nodes. Histopathology 1994;25:483. 15. Domagala W, Bedner E, Chosia M, et al. Keratin-positive reticulum cells in fine needle aspirates, and touch imprints of hyperplastic lymph nodes. Acta Cytol 1991;36:241 5. INVITED COMMENTARY This article from the Mitsui Memorial Hospital once again identifies the fact that micrometastasis to hilar and mediastinal lymph nodes can be identified by immunohistochemical staining when routine pathologic examination fails to identify metastatic disease. In their methodology, they do not indicate how many routine sections from each lymph node block were examined by hematoxylin and eosin staining and whether further sections would have altered the original No result. Despite this, their findings confirm the findings of many other groups. Much more intriguing is the conclusion reached that an extended mediastinal lymph node dissection to include the superior mediastinal and thoracic inlet region in left-sided tumors may improve survival by removing all sites of occult micrometastatic involvement. Dr Hata s group has performed extended mediastinal lymphadenectomy for left-sided tumors for almost 20 years and their results have convinced other Japanese surgeons to modify their approach to left-sided tumors. Unfortunately, their experience has never been reported in the English literature and, indeed, has only been reported in abstract form. In their abstract presentations, they have claimed that extended mediastinal lymphadenectomy improves survival in left-sided tumors. Recently, Keller and colleagues [1] analyzed the result of a North American trial, comparing patients undergoing mediastinal lymph node dissection with those receiving nodal sampling only. In left-sided tumors, where the superior mediastinal lymph nodes are not routinely dissected, there was no difference in results between sampling and lymphadenectomy. However, in right-sided tumors, where the superior mediastinal lymph nodes are routinely dissected during lymphadenectomy, the 5-year survival was improved by this latter procedure. Kellar and colleagues suggested that the equivalent survival between sampling and lymphadenectomy on the left side perhaps was the failure of left-sided lymphadenectomy to remove the superior mediastinal lymph node tissue. Two current ACOSOG trials (ACOSOG Z0030 and Z0040) may shed further light on all of these questions as, in those trials, superior mediastinal lymphadenectomy is not carried out in left-sided tumors and an attempt is being made to identify occult mediastinal nodal disease using immunohistochemistry. Whether or not routine mediastinal lymphadenectomy improves survival in lung cancer is still a moot point and is worthy of further study. I would urge the Mitsui Memorial Hospital group to publish their total experience with extended lymphadenectomy for left-sided tumors in a widely read journal so that the results of this approach can be analyzed by all interested parties. Robert J. Ginsberg, MD Division of Thoracic Surgery Toronto General Hospital 200 Elizabeth Street, EN-10-226 Toronto, Ontario M56 2C4, Canada e-mail: robert.ginsberg@uhn.on.ca. Reference 1. Keller SM, Adak S, Wagner H, et al. Complete mediastinal lymph node dissection improve survival in patients with resected stage II and IIa non-small cell lung cancer. Ann Thorac Surg 2000;70:358 66. 2002 by The Society of Thoracic Surgeons 0003-4975/02/$22.00 Published by Elsevier Science Inc PII S0003-4975(02)03398-2