Lymphovascular Invasion in Non–Small-Cell Lung Cancer Implications for Staging and Adjuvant Therapy

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector ORIGINAL ARTICLE

Kristin A Higgins, MD,* Junzo P Chino, MD,† Neal Ready, MD,‡ Thomas A D’Amico, MD,§ Mark F Berry, MD,§ Thomas Sporn, MD,|| Jessamy Boyd, MD,# and Chris R Kelsey MD†

tage is currently the dominant factor affecting adjuvant- Background: Lymphovascular space invasion (LVI) is an established treatment recommendations in resected non–small-cell negative prognostic factor and an indication for postoperative radia- S lung cancer (NSCLC).1 Practice guidelines have been pro- tion therapy in many malignancies. The purpose of this study was to foundly influenced by three recent randomized trials dem- evaluate LVI in patients with early-stage non–small-cell lung cancer, onstrating a survival benefit with the addition of adjuvant undergoing surgical resection. chemotherapy.2–4 This benefit has been largely confined to N1 Methods: All patients who underwent initial surgery for pT1-3N0-2 to N2 disease,5 although subset analyses have suggested that non–small-cell lung cancer at Duke University Medical Center from chemotherapy is also advantageous for patients with larger 1995 to 2008 were identified. A multivariate ordinal regression was stage I tumors.4,6 The role of postoperative radiation therapy is used to assess the relationship between LVI and pathologic hilar and/ less well defined, but is generally recommended for patients or mediastinal lymph node (LN) involvement. A multivariate Cox with involved mediastinal lymph nodes (LNs) (N2) or positive regression analysis was used to evaluate the relationship of LVI and surgical margins. other clinical and pathologic factors on local failure (LF), freedom Elucidation of high-risk pathologic features in resected from distant metastasis (FFDM), and overall survival (OS). Kaplan- NSCLC may help to further stratify patients into risk groups, Meier methods were used to generate estimates of LF, FFDM, and allowing for further refinement of adjuvant treatment recom- OS in patients with and without LVI. mendations. For example, lymphovascular space invasion Results: One thousand five hundred and fifty-nine patients were iden- (LVI) is an adverse prognostic factor for numerous solid tified. LVI was independently associated with the presence of regional tumors. Furthermore, the presence of LVI is an independent LN involvement (p < 0.001) along with lobar (versus sublobar) resec- indication for postoperative radiation therapy in several epi- tions (p < 0.001), and an open thoracotomy (versus video-assisted thelial malignancies, including cervix,7 endometrial,8 vulvar,9 thoracoscopic surgery). LVI was not independently associated with and head and neck cancers.10 LF on multivariate analysis (hazard ratio [HR] = 1.23, p = 0.25), but Several studies have demonstrated that LVI is also a was associated with a lower FFDM (HR 1.52, p = 0.005) and OS poor prognostic factor for recurrence-free and overall survival (HR 1.26, p = 0.015). In addition, multivariate analysis showed that (OS) in NSCLC.11–15 One recent study has demonstrated that LVI was strongly associated with increased risk of developing distant presence of LVI in a biopsy specimen is an independent risk metastases (HR = 1.75, p = 0.006) and death (HR = 1.53, p = 0.003) factor for death and metastatic progression for patients treated in adenocarcinomas but not in squamous carcinomas. with definitive chemoradiotherapy.16 However, the influence of Conclusions: LVI is associated with an increased risk of harboring LVI on adjuvant-treatment recommendations after resection is regional LN involvement. LVI is also an adverse prognostic factor for unclear. Current NCCN guidelines are somewhat ambiguous, the development of distant metastases and long-term survival. stating that the presence of LVI is considered “high-risk”.17 Key Words: Lymphovascular invasion, Lung cancer, Local failure. Whether LVI increases the risk of local or distant recurrence (or both) is unknown. In addition to this question, we also (J Thorac Oncol. 2012;7: 1141–1147) sought to investigate the association between the presence of LVI in the primary tumor and the risk of harboring disease in hilar and/or mediastinal LNs. *Department of Radiation Oncology, Emory University, Atlanta, Georgia; †Department of Radiation Oncology, ‡Department of Medicine, Division of Medical Oncology, §Department of Surgery, Division of Cardiovascular METHODS and Thoracic Surgery, and ||Department of Pathology, Duke University This Institutional Review Board-approved study identi- Medical Center, Durham North Carolina; and #National Comprehensive Cancer Network, US Oncology, Dallas, Texas. fied all patients who underwent surgery for T1-3N0-2 NSCLC Disclosure: The authors declare no conflicts of interest. at Duke University between 1995 and 2008. Patients were Address for correspondence: Kristin A. Higgins, MD, Department of excluded if they received preoperative chemotherapy and/or Radiation Oncology, Emory University, Winship Cancer Institute, Atlanta, preoperative radiation therapy, or had a prior history of lung GA 30322. E-mail: [email protected] cancer. Medical records and pertinent radiological imaging Copyright © 2012 by the International Association for the Study of Lung Cancer were reviewed to characterize each patient’s demographic ISSN: 1556-0864/12/0707-1141 information, obtain surgical and pathological details, and

Journal of Thoracic Oncology • Volume 7, Number 7, July 2012 1141 Higgins et al. Journal of Thoracic Oncology • Volume 7, Number 7, July 2012

score patterns of failure after surgery. The pathologic stag- intervals, and comparisons were made via the log-rank test. ing of patients in this cohort reflects the updated tumor, node, When generating survival curves for patients with LVI versus metastasis staging for NSCLC.1 no LVI, the patients with unknown LVI status were grouped Between 1995 and 1997, lung tumor specimens were into the “no LVI” category. Progression-free survival was cal- interpreted by multiple pathologists at Duke University culated from the date of surgery to date of treatment failure Medical Center. After 1998, all specimens were interpreted (defined as local and/or distant recurrence, development of a by a single pathologist with a special interest in pulmonary second primary lung cancer, or death from any cause). Time pathology (TAS). Hematoxylin and eosin stains were per- to LF and time to distant failure (DF) were calculated from the formed on all specimens. LVI was reported when tumor cells date of surgery to date of local or distant recurrence, respec- were demonstrated on histologic examination within lym- tively. For these analyses, patients developing what was felt to phatic channels, veins/venules, and/or arteries/arterioles. be a second primary lung cancer were censored on the date the Patterns of failure were assessed by means of follow-up second primary malignancy was diagnosed. Local and distant imaging studies and information obtained from procedures such recurrences were scored independently (i.e., patients develop- as computed-tomography (CT)–guided transthoracic biopsies, ing a distant recurrence were not censored for LF, but were bronchoscopy, endobronchial ultrasound, and mediastinoscopy, assessed for LF until the date of last follow-up or death). etc. A local failure (LF) was scored when disease recurred at the surgical resection margin or regional LNs (ipsilateral hilum and/or mediastinum). Nodal failures in the ipsilateral hilum or RESULTS mediastinum were defined as a new or enlarging LN of 1 cm or Between 1995 and 2008, 1559 patients met the inclu- more on short axis on CT, and/or hypermetabolic on positron sion criteria. Median follow-up for all patients was 34 months emission tomography imaging, which in the patient’s subse- and for survivors it was 40 months. Patient characteristics are quent clinical follow-up was consistent with disease progres- reported in Table 1. The majority (98%) had hilar and/or medi- sion. All cases of possible local and distant recurrence were astinal LN sampling/dissection at the time of surgery. The reviewed by two authors (CK and JB/KH) to ensure accuracy. majority of patients were stage I, with 41% of patients with All patients had routine postsurgical surveillance with imaging pathologic stage IA disease and 26% of patients with stage IB studies, including chest CT, but the frequency and choice of disease. Presence of LVI was noted in 23% of the patients. imaging modality was not standardized. LVI and Regional LN Involvement Statistical Analysis Patients with no regional LN sampling were excluded from this analysis (n = 33). Of the remaining 1526 patients, The Mann-Whitney U test was used to test for an asso- 1219 were pathologic N0, 235 patients were pathologic N1, ciation between LVI and the presence of involved regional and 72 patients were pathologic N2. LNs. Primary tumor size, type of surgery (wedge/segment- Factors associated with regional LN involvement on uni- ectomy versus lobectomy/pneumonectomy), thoractomy ver- variate analysis included LVI, increasing size of the primary sus a video-assisted thoracoscopic surgical (VATS) approach, tumor, lobar versus sublobar resection, thoracotomy versus histology, visceral pleural invasion, grade, and primary tumor VATS approach, visceral pleural invasion, and squamous/large- location (right versus left, upper/middle versus lower) were cell histology. On multivariate ordinal regression, the pres- also included in this assessment. Factors with a p value of ence of LVI (p < 0.001), thoracotomy (p < 0.001), and lobar 0.1 or less were included in a multivariate ordinal regres- resections (p < 0.001) continued to be significant predictors of sion. Patients without regional LN sampling/dissection were harboring pathologically involved regional LNs (Table 2). excluded from this part of the analysis. In patients who underwent lobectomy or pneumonec- Univariate and multivariate analyses were also per- tomy (n = 696), the risk of harboring ipsilateral hilar LN dis- formed to examine clinical and pathological factors associated ease was 36% versus 21% (p < 0.001) and the risk of harboring with LF, development of distant metastases (DM), and OS, mediastinal LN disease was 12% versus 5% (p < 0.001), for including LVI, type of surgical resection (thoracotomy versus patients with and without LVI, respectively. Furthermore, VATS), size of primary tumor, number of hilar LNs sampled, LVI was associated with increasing number of positive LNs number of hilar LNs involved, number of mediastinal LN sta- (p < 0.001), with the median number of positive LNs equal to tions sampled, number of mediastinal LNs involved, tumor one (interquartile range 1–2) for patients without LVI and two 18 tumor grade, histology (squamous/large-cell versus others), (interquartile range 1–4) for patients with LVI. visceral pleural invasion, stage, adjuvant chemotherapy, age, and sex. Factors with a p value of 0.1 or less on univariate analysis were included in the multivariate model. For the LF LVI and Risk of Local Recurrence analysis, patients with positive margins and/or those receiving For this analysis, patients with positive surgical margins postoperative radiation were excluded. Univariate and multi- and those receiving postoperative radiation were excluded, variate analyses were also performed based on histology, with leaving 1458 evaluable patients. On univariate analysis, LVI LF, DM, and OS examined separately in squamous-cell carci- was associated with statistically significant increased risk of nomas and adenocarcinomas. LF with Kaplan-Meier estimates of 5-year local control of The Kaplan-Meier product-limit method19,20 was used 79% versus 71% (p = 0.001) (Fig. 1). However, on multivari- to estimate 5-year recurrence probabilities and confidence ate analysis this association was lost (Table 4).

TABLE 1. Patient Characteristics TABLE 2. Factors Associated with Regional Lymph Node Involvement Characteristic a Age (yrs) Factors p Value Median 67 Lobectomy versus wedge/ segmentectomy <0.01 Range 20–91 Lymphovascular invasion <0.01 Race Tumor size 0.094 African-American 202 (13%) Squamous/large cell versus adenocarcinoma 0.80 White 1328 (85%) Moderate/high grade versus low grade 0.12 Other 29 (2%) Open resection versus VATS approach <0.01 Sex Visceral pleural invasion 0.08 Male 833 (53%) aMultivariate ordinal regression analysis. Female 726 (47%) VATS, video-assisted thoracoscopic surgical. Pathologic stage (TNM 7) IA 638 (41%) IB 411 (26%) IIA 272 (17%) LVI and DM, Recurrence-Free Survival, and OS IIB 124 (10%) The presence of LVI was found to be significantly asso- IIIA 112 (7%) ciated with a detriment in freedom from DM as depicted in IIIB 2 (0.1%) Figure 2 (57% versus 73% at 5 years, p < 0.001). This remained Histology significant on multivariate analysis (Table 4). Adenocarcinoma 734 (47%) The presence of LVI was also associated with increased Squamous cell 561 (36%) risk of death (5-year OS 31% versus 56%, p < 0.001) Large cell 93 (6%) (Fig. 3), which remained significant on multivariate analysis Bronchioloalveolar 40 (3%) (Table 4). Adjuvant chemotherapy was also associated with Adenosquamous 21 (1%) improved OS on multivariate analysis (p < 0.001), and seemed NSCLC NOS 110 (7%) to improve the 5-year survival for patients with LVI (32% ver- Surgical Approach sus 60%, p = 0.001) and for patients without LVI (49% versus 69%, p < 0.001). In addition, when examined by histology, Open 795 (51%) LVI was strongly associated with increased risk of developing VAT S 764 (49%) DM (hazard ratio [HR] = 1.75, p = 0.006) and death (HR = Surgical Procedure 1.53, p = 0.003) on multivariate analysis in adenocarcinomas. Wedge/Segmentectomy 305 (20%) Multivariate analysis did not find LVI to be statistically sig- Lobectomya 1154 (74%) nificant for DM (HR = 1.08, p = 0.75) or death (HR = 1.23, Pneumonectomy 100 (6%) p = 0.11) in squamous-cell carcinoma. Mediastinal sampling/dissection Ye s 1361 (87%) No 198 (13%) DISCUSSION Lymphovascular space invasion LVI is an adverse prognostic factor in many other solid Ye s 352 (23%) tumors and its presence is an independent indication for post- No 1107 (71%) operative radiotherapy in head and neck and multiple gyneco- Unknown 100 (6%) logical malignancies. The presence of LVI does not currently Visceral Pleural Invasion influence treatment recommendations in NSCLC. Ye s 366 (24%) In the present study, the presence of LVI within the pri- No 1128 (72%) mary tumor was independently associated with a higher risk Unknown 65 (4%) that regional LNs were pathologically involved. Such infor- Chemotherapy mation could be useful when determining treatment recom- Ye s 184 (12%) mendations for patients who do not undergo regional LN No 1375 (88%) sampling/dissection at the time of surgery or in patients without LN sampling before undergoing definitive radiotherapy. NSCLC NOS, non–small-cell carcinoma not otherwise specified; VATS, video- assisted thoracoscopic surgery; TNM, tumor, node, metastasis. Not surprisingly, increased risk of regional node involvement aIncluding bilobectomy and sleeve lobectomy. was also associated with lobectomy compared with sublobar resection, likely because of the absence of LN dissection or a limited number of LNs retrieved with most sublobar resections. In addition, open thoracotomy was also associated with When examining the effect of LVI on a stage-by-stage increased risk of regional node involvement, perhaps because basis, a statistically significant detriment in local control was seen patients not eligible for a VATS procedure had a greater extent in patients with stage IB disease with LVI versus stage IB without of the disease. LVI (61% versus 81%, p < 0.01). Presence of LVI was not statisti- The therapy of NSCLC is now being driven by histology, cally significant for LF in the other stage groups (Table 3). along with the presence of validated predictive biomarkers

FIGURE 1. Local control based on the presence or absence of LVI. LVI, lymphovascular space invasion.

TABLE 3. Local Control Stage by Stage Stratified by Lymphovascular Invasion 5-Year Local Stage Control (%) 95% CI (%) p Value IA 86 82–90 0.43 Lymphovascular invasion 92 85–99 No lymphovascular invasion 85 81–89% IB 77 71–84 <0.01 Lymphovascular invasion 61 43–79 No lymphovascular invasion 81 74–87 IIA 63 54–72 0.22 Lymphovascular invasion 59 44–74 No lymphovascular invasion 65 54–76 IIB 63 48–77 0.45 Lymphovascular invasion 60 32–89 No lymphovascular invasion 62 46–79 IIIA 69 51–86 0.94 Lymphovascular invasion 67 38–97 No lymphovascular invasion 70 48–91

CI, confidence interval.

FIGURE 2. Freedom from distant metastasis based on the presence or absence of LVI. LVI, lymphovascular space invasion.

TABLE 4. Multivariate Analyses for Local Recurrence, Distant Metastases, and Overall Survival Local Failure Distant Metastases Overall Survival Hazard Hazard Hazard Factorsa Ratio 95% CI p Value Ratio 95% CI p Value Ratio 95% CI p Value Surgical procedure Wedge/ segmentectomy 1.78 1.23–2.58 <0.01 1.56 1.30–1.88 <0.01 ≥Lobectomy Lymphovascular invasion Ye s 1.23 0.87–1.76 0.25 1.52 1.13–2.02 <0.01 1.25 1.05–1.51 0.02 No Age (per yr) 0.99 0.97–0.99 0.03 1.02 1.01–1.03 <0.01 Tumor size (per cm) 1.13 1.05–1.22 <0.01 1.16 1.10–1.23 <0.01 1.08 1.04–1.13 <0.01 Visceral pleural invasion Ye s 2.07 1.48–2.89 <0.01 1.30 0.96–1.76 0.09 1.12 0.93–1.36 0.24 No Grade Moderate/high 1.66 0.86–3.21 0.13 1.42 0.86–2.35 0.17 1.33 1.00–1.77 0.05 Low Histology Squamous or large cell 1.48 1.07–2.04 0.02 1.13 0.96–1.32 0.15 Non-squamous VATS approach Ye s 0.91 0.65–0.91 <0.01 0.84 0.63–1.11 0.22 0.89 0.75–1.05 0.16 No (Continued)

TABLE 4. (Continued) Local Failure Distant Metastases Overall Survival Hazard Hazard Hazard Factorsa Ratio 95% CI p Value Ratio 95% CI p Value Ratio 95% CI p Value Adjuvant chemotherapy Ye s 0.66 0.41–1.06 0.09 0.67 0.46–0.98 0.04 0.35 0.24–0.50 <0.01 No Male sex Ye s 1.13 0.83–1.55 0.44 1.32 1.13–1.55 <0.01 No Number of N1 LNs sampled (per node) 0.99 0.96–1.03 0.69 Number of involved N1 LNs (per node) 1.12 0.98–1.29 0.10 1.12 1.00–1.25 0.05 1.12 1.04–1.21 <0.01 Number of involved N2 LNs (per node) 1.31 0.91–1.88 0.15 1.37 1.06–1.77 0.02 1.31 1.09–1.57 <0.01

aOnly factors with a p value less than 0.1 on univariate analysis were included in the multivariate model. The second factor listed is the reference set. LN, lymph node; CI, confidence interval; VATS, video-assisted thoracoscopic surgical.

FIGURE 3. Overall survival based on the presence or absence of LVI. LVI, lymphovascular space invasion.

for erlotinib and crizotinib.21,22 Interestingly, when patients were 3. Arriagada R, Bergman B, Dunant A, Le Chevalier T, Pignon JP, analyzed according to histology, the association of LVI with Vansteenkiste J; International Adjuvant Lung Cancer Trial Collaborative Group. Cisplatin-based adjuvant chemotherapy in patients with completely increased risk of DM and death was only found in adenocarcino- resected non-small-cell lung cancer. N Engl J Med 2004;350:351–360. mas. Currently, the presence of LVI is rarely reported in pathol- 4. Winton T, Livingston R, Johnson D, et al.; National Cancer Institute of ogy reports from core needle biopsies of lung masses, as tumor Canada Clinical Trials Group; National Cancer Institute of the United stroma containing vessels and lymphatics must be present to make States Intergroup JBR.10 Trial Investigators. Vinorelbine plus cisplatin such a determination. Improving biopsy techniques to allow for vs. observation in resected non-small-cell lung cancer. N Engl J Med 2005;352:2589–2597. more adequate tumor sampling is necessary to routinely report 5. Pignon JP, Tribodet H, Scagliotti GV, et al.; LACE Collaborative Group. presence or absence of LVI and determine tumor histology and Lung adjuvant cisplatin evaluation: a pooled analysis by the LACE molecular status. Although the present study demonstrates LVI to Collaborative Group. J Clin Oncol 2008;26:3552–3559. be a poor prognostic factor in resected lung specimens, a recently 6. Strauss GM, Herndon JE 2nd, Maddaus MA, et al. Adjuvant paclitaxel published study has also demonstrated that LVI in core needle plus carboplatin compared with observation in stage IB non-small-cell biopsies is associated with poor outcome in patients undergoing lung cancer: CALGB 9633 with the Cancer and Leukemia Group B, 16 Radiation Therapy Oncology Group, and North Central Cancer Treatment chemoradiation. We believe that refining tumor characterization Group Study Groups. J Clin Oncol 2008;26:5043–5051. at core needle biopsy is an important step forward. 7. Sedlis A, Bundy BN, Rotman MZ, Lentz SS, Muderspach LI, Zaino RJ. The presence of LVI was associated with a higher risk of LF A randomized trial of pelvic radiation therapy versus no further therapy on univariate analysis. This distinction was most evident in stage in selected patients with stage IB carcinoma of the cervix after radical IB patients (i.e., those with larger primary tumors but no evidence hysterectomy and pelvic lymphadenectomy: A Gynecologic Oncology Group Study. Gynecol Oncol 1999;73:177–183. of regional LN involvement). In this patient population, LVI has 8. Keys HM, Roberts JA, Brunetto VL, et al.; Gynecologic Oncology Group. been associated with inferior OS in multiple studies,11,14 but the A phase III trial of surgery with or without adjunctive external pelvic impact of LVI on LF was not evaluated. Although significant on radiation therapy in intermediate risk endometrial adenocarcinoma: a univariate analysis, LVI did not remain statistically significant for Gynecologic Oncology Group study. Gynecol Oncol 2004;92:744–751. LF for the entire cohort on multivariate analysis. In contrast, other 9. Heaps JM, Fu YS, Montz FJ, Hacker NF, Berek JS. Surgical-pathologic variables predictive of local recurrence in squamous cell carcinoma of the pathologic factors were strongly associated with LF, including vulva. Gynecol Oncol 1990;38:309–314. visceral pleural invasion and squamous/large-cell histology. 10. Huang TY, Hsu LP, Wen YH, et al. Predictors of locoregional recurrence Our data indicate that LVI is more strongly associ- in early stage oral cavity cancer with free surgical margins. Oral Oncol ated with DF as opposed to LF. Patients with LVI had sig- 2010;46:49–55. nificantly increased risk of developing DM and a lower rate 11. Hsu CP, Hsia JY, Chang GC, et al. Surgical-pathologic factors affect long- term outcomes in stage IB (pT2 N0 M0) non-small cell lung cancer: a of survival. Other studies have demonstrated an association heterogeneous disease. J Thorac Cardiovasc Surg 2009;138:426–433. between LVI and decreased recurrence-free survival and OS 12. Giraud P, Antoine M, Larrouy A, et al. Evaluation of microscopic in NSCLC.12–14 The present study is the first to evaluate LF tumor extension in non-small-cell lung cancer for three-dimensional and DF as distinct entities in association with LVI. Why LVI conformal radiotherapy planning. Int J Radiat Oncol Biol Phys is more strongly associated with DF and death in adenocarci- 2000;48:1015–1024. 13. Krauss S, Perez C, Lowenbraun S, Sonoda T, Bartolucci A, Buchanan R. nomas is unknown. Combined modality treatment of localized small-cell lung carcinoma. Limitations to the present study do exist. First, given its ret- A randomized prospective study of the Southeastern Cancer Study Group. rospective nature, no centralized confirmatory pathologic review Cancer Clin Trials 1980;3:297–305. was possible. However, the vast majority of specimens were 14. Meyskens FL Jr, Berglund EB, Saxe DF, et al. Biological and biochemical analyzed by a single pathologist who specializes in pulmonary properties of a human uveal melanocyte-derived cell line. In Vitro pathology. Second, lymphatic invasion and vascular invasion 1980;16:775–780. 15. Shiono S, Abiko M, Sato T. Positron emission tomography/computed were analyzed together and not independent of one another. It is tomography and lymphovascular invasion predict recurrence in stage I not known if one is of greater importance than the other. Finally, lung cancers. J Thorac Oncol 2011;6:43–47. although the sample cohort was large, the majority of patients 16. Massabeau C, Filleron T, Wakil G, et al. The prognostic significance of had stage I disease. Statistical power may be limited in stage II lymphovascular invasion on biopsy specimens in lung cancer treated with and III patients, the subset of patients with the highest risk of LF. definitive chemoradiotherapy. Clin Lung Cancer 2012;13:59–67. 17. Ettinger DS, Akerley W, Borghaei H, et al. NCCN Clinical Practice In conclusion, the presence of LVI is associated with an Guidelines in Oncology, Non-Small Cell Lung Cancer, (ed Version increased risk of developing DM and death after surgery for 2.2012). Fort Washington, PA: National Comprehensive Cancer Network, early-stage NSCLC. LVI alone does not seem to be a strong 2012. enough risk factor for LF alone to warrant postoperative radi- 18. Kelsey CR, Marks LB, Hollis D, et al. Local recurrence after surgery for ation therapy in NSCLC. The presence of LVI is associated early stage lung cancer: an 11-year experience with 975 patients. Cancer 2009;115:5218–5227. with an increased risk of developing DM and may be helpful 19. Collett D. Modeling Survival Data in Medical Research, 2 nd Ed. London: in the decision paradigm for adjuvant chemotherapy. Chapman & Hall, 2003. 20. Therneau TM, Grambsch PM. Modeling Survival Data. New York: Springer, 2000. REFERENCES 21. Kwak EL, Bang YJ, Camidge DR, et al. Anaplastic lymphoma 1. AJCC. Cancer Staging Manual- Lung ,7 th Ed. New York, Springer, 2010. kinase inhibition in non-small-cell lung cancer. N Engl J Med Pp. 253–270. 2010;363:1693–1703. 2. Douillard JY, Rosell R, De Lena M, et al. Adjuvant vinorelbine plus cisplatin 22. Pao W, Miller V, Zakowski M, et al. EGF receptor gene mutations are versus observation in patients with completely resected stage IB-IIIA non- common in lung cancers from “never smokers” and are associated with small-cell lung cancer (Adjuvant Navelbine International Trialist Association sensitivity of tumors to gefitinib and erlotinib. Proc Natl Acad Sci USA [ANITA]): a randomised controlled trial. Lancet Oncol 2006;7:719–727. 2004;101:13306–13311.