HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to Personal Folders Download to citation manager Author home page(s): A. Pieter Kappetein Michel I.M. Versteegh Ad J.J.C. Bogers Permission Requests Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Birim, O. Articles by Bogers, A. J.J.C. Search for Related Content PubMed PubMed Citation Articles by Birim, O. Articles by Bogers, A. J.J.C. Related Collections Lung - cancer

J Thorac Cardiovasc Surg 2006;132:491-498
© 2006 The American Association for Thoracic Surgery

General Thoracic Surgery

Long-term survival after non–small cell lung cancer surgery: Development and validation of a prognostic model with a preoperative and postoperative mode

^a Department of Cardio-Thoracic Surgery, Erasmus MC, Rotterdam, The Netherlands
^c Department of Public Health, Erasmus MC, Rotterdam, The Netherlands
^b Department of General Surgery, Leiden University Medical Center, Leiden, The Netherlands
^d Department of Cardio-Thoracic Surgery, Leiden University Medical Center, Leiden, The Netherlands

Received for publication December 2, 2005; revisions received February 26, 2006; accepted for publication April 11, 2006.

^_* Address for correspondence: Özcan Birim, MD, PhD, Department of Cardio-Thoracic Surgery, Room BD 156, Erasmus MC, P.O. Box 2040, 3000 CA Rotterdam, The Netherlands (Email: o.birim{at}erasmusmc.nl).

	Abstract

Top Abstract Introduction Methods Results Discussion References

 
OBJECTIVE: At present, there is no prognostic model that is specific for prediction of survival after non–small cell lung cancer surgery. We aimed to develop a prognostic model that can be used to estimate the postoperative survival of individual patients.

METHODS: A total of 766 patients underwent resection for primary non–small cell lung cancer. Comorbid conditions were scaled according to the Charlson comorbidity index (CCI). Cox proportional hazard analyses were used to determine risk factors for survival. A prognostic model for survival with a preoperative and postoperative mode was established. Performance of the prognostic model, the CCI, and pathologic tumor stage were quantified by a concordance statistic to indicate discriminative ability.

RESULTS: The factors associated with an impaired survival were male sex, age, chronic obstructive pulmonary disease, congestive heart failure, any prior tumor, moderate-to-severe renal disease (preoperative and postoperative mode), clinical tumor stage (preoperative mode), type of resection, and pathologic tumor stage (postoperative mode). The discriminative performance was poor for the CCI (c = 0.55), better for pathologic tumor stage (c = 0.60) and for the preoperative mode (c = 0.61), and best for the postoperative mode (c = 0.65). The discriminative performance of the postoperative mode was better than the discriminative performance of the CCI (P < .0001), the preoperative mode (P < .0002), and pathologic tumor stage (P < .0001). The discriminative performance of the preoperative mode was better than the discriminative performance of the CCI (P < .0001) and similar (P = .90) to a model that only included pathologic tumor stage.

CONCLUSIONS: The prognostic model, particularly the postoperative mode, successfully estimates long-term survival of individual patients and could help clinicians in clinical decision-making and treatment tailoring.

	Introduction

Top Abstract Introduction Methods Results Discussion References

 
Surgical resection is generally the mainstay of curative treatment in early-stage non–small cell lung cancer (NSCLC). Survival is mainly dependent on pathologic tumor stage, with a 5-year survival that ranges between 67% and 38% for pathologic tumor stage IA to IIB.^1,2 However, many other factors also have an impact on survival, such as comorbidity, age, and type of resection.^3-6 Owing to the increasing age of the population diagnosed with NSCLC, the incidence of comorbidity and consequently high-risk patients also increases. The most frequent comorbidities in patients with NSCLC are cardiovascular disease (23%), chronic obstructive pulmonary disease (COPD) (22%), and other malignancies (15%).⁷

Several scoring systems based on various comorbid conditions have previously been used to stratify patients according to risk of complications and long-term survival after NSCLC surgery.^8-11 In previously published studies we^8,9 found the Charlson comorbidity index (CCI) to predict postoperative outcome more accurately than the individual comorbid conditions. However, these scoring systems are developed in other patient populations and thus may be suboptimal for application in patients having NSCLC surgery. In addition, these models do not include several other prognostic factors in patients having NSCLC surgery, such as sex, age, extent of resection, and tumor stage. Furthermore, these models do not estimate the long-term survival of individual patients after surgical resection. At present, there is no accepted prognostic model that is specific for NSCLC patients and one that can be used to estimate the long-term survival of individual patients after surgical resection of NSCLC.

In addition, to provide surgeons and patients with better quality information on risk assessment and postoperative survival, we performed this retrospective study. We aimed to identify prognostic factors for survival in NSCLC surgery, to develop a prognostic model with a preoperative and postoperative mode to estimate survival of individual patients, and to compare the predictive accuracy of the prognostic model with the predictive accuracy of the CCI and pathologic tumor stage.

	Methods

Top Abstract Introduction Methods Results Discussion References

 
Patients
We retrospectively reviewed the medical records of 766 consecutive patients who underwent resection for primary NSCLC at the Departments of Cardio-Thoracic Surgery of the Erasmus MC Rotterdam (433 patients) and the Leiden University Medical Center (333 patients), from January 1, 1989, through December 31, 2001. Patients were followed up with regular visits to the outpatient clinic. Civil administrations were consulted to assess late mortality. Follow-up was completed through August 2004. Survival time was defined as the difference between the date of surgery and the date of last follow-up. Hospital mortality was defined as death occurring within 30 days of surgery or any death later during the same postoperative hospital stay.

In all patients, preoperative diagnostic workup included a complete medical history, physical examination, plain chest radiography, electrocardiography, routine laboratory tests, lung function tests, and computed tomography of the chest and upper abdomen. Additional staging procedures, namely, mediastinoscopy and liver, bone, and brain scans were selectively performed to aid in treatment planning according to best clinical practice at the time of presentation. In retrospect, each patient was assessed for preoperative CCI.^8,9 The index can be divided into four comorbidity grades: 0, 1 to 2, 3 to 4, and 5 or more.

COPD as a comorbid condition is defined according to the GOLD criteria (Global Initiative for Chronic Obstructive Lung Disease) as a postbronchodilator forced expiratory volume in 1 second (FEV₁)/forced vital capacity ratio less than 70%.¹²

Histologic typing occurred according to the World Health Organization Histologic Typing of Lung Tumours.¹³ Clinical and pathologic tumor staging of the patients occurred according to the international TNM classification for lung cancer.¹ Staging was based on pathologic assessment of the primary tumor, and lymph node assessment was carried out with preoperative mediastinoscopy (clinical–pathologic staging) or surgical sampling of bronchopulmonary, hilar, and mediastinal lymph nodes (pathologic staging).

The following risk factors for survival were evaluated: sex, age, type of resection, histologic cell type, smoking, COPD (unknown in 57 patients), FEV₁% (unknown in 46 patients), clinical tumor stage, pathologic tumor stage, neoadjuvant therapy, adjuvant therapy, and each common condition of the CCI.

Statistical Analysis and Development of the Prognostic Model
Discrete variables are displayed as proportions, and continuous variables are displayed as means ± standard deviations unless specified otherwise. We used the Cox proportional hazard analysis to determine risk factors for survival, where effects were expressed as relative risks with 95% confidence intervals (CI). Multivariate analysis was performed with stepwise backward elimination of variables, starting with a model in which each variable with a P value of less than .20 in the univariate analysis was entered. Because COPD, FEV₁%, and chronic pulmonary disease were correlated to each other, we entered only the most significant factor (COPD) in the multivariate analysis. In the multivariate analysis the estimated mean value of COPD was imputed for the 57 missing values.

Preoperative and a postoperative modes of the prognostic model for survival were established. This model considered sex and age and included those factors that were available preoperatively or postoperatively that were associated with an impaired survival in the multivariate analysis. Factors that were less important but almost significant, such as COPD, congestive heart failure, and left-sided pneumonectomy, were retained in the model because of their documented relevance in the literature.^5,14-16 The performance of the prognostic model, the CCI, and pathologic tumor stage were quantified by a concordance statistic (c statistic), which is similar to the area under the receiver operating characteristic curve for binary data.¹⁷ A c statistic of 0.5 indicates that the model has no discriminative ability, and a c statistic of 1 indicates that the model perfectly distinguishes between those who die early and those who die later.

Bootstrapping techniques were used for internal validation of the c statistic of the models.^14,18 Bootstrap samples were drawn with replacement and with the same size as the original sample. Regression models were created in each bootstrap sample and tested on the original sample. This procedure was repeated 200 times to obtain stable estimates of the optimism of the model, that is, how much the model performance was expected to decrease when applied in future patients.¹⁹

Bootstrap resampling was also used to test differences in performance between alternative models. The mean difference and standard error was estimated from 2000 bootstraps.

For practical application, we developed an Excel spreadsheet in which preoperative and postoperative characteristics can be entered and the predicted survival with 95% CI is automatically calculated. This Excel spreadsheet for easy access and use by clinicians is available on the Internet (http://www.cardiothoracicresearch.nl). Statistical calculations were performed with SPSS (version 12.0; SPSS Inc, Chicago, Ill) and S-plus (version 6.0; Insightful Corp, Seattle, Wash).

	Results

Top Abstract Introduction Methods Results Discussion References

 
Of the 766 patients enrolled in this analysis, 600 (78%) were men and 166 (22%) were women. Median follow-up time was 3.5 ± 0.3 years. The mean age at operation was 64.5 ± 9.2 years (range, 37-84 years). The patients' preoperative characteristics are outlined in Tables 1 and 2. The types of procedures performed consisted of a wedge resection (17; 2%), a lobectomy (470; 61%), a bilobectomy (73; 10%), a left-sided pneumonectomy (105; 14%), and a right-sided pneumonectomy (101; 13%). Thirty-four (4%) patients received neoadjuvant chemotherapy, no patient was treated with neoadjuvant radiotherapy, 107 (14%) patients were treated with adjuvant radiotherapy, and 3 (0%) patients were treated with postoperative chemotherapy. Tumors were classified histologically as squamous cell carcinoma (386; 50%), adenocarcinoma (246; 32%), large cell carcinoma (82; 11%), and bronchoalveolar cell carcinoma (52; 7%). The patients' operative demographics are listed in Table 3.

Predictors for impaired survival in the univariate analysis included male sex, age, COPD, FEV₁% less than 70, type of resection, CCI score of 3 or more, congestive heart failure, chronic pulmonary disease, any prior tumor within 5 years of diagnosis, moderate-to-severe renal disease, leukemia, lymphoma, clinical tumor stage, and pathologic tumor stage (Table 4). The factors associated with an impaired survival in the multivariate analysis were male sex, age, COPD, type of resection, congestive heart failure, any prior tumor within 5 years of diagnosis, moderate-to-severe renal disease, clinical tumor stage, and pathologic tumor stage (Table 4). The developed Excel spreadsheet of the preoperative mode and postoperative mode of the model is illustrated in Figures 1 and 2, respectively. In Figure 1, the clinical factors of a 64-year old man with clinical tumor stage IB NSCLC are filled in and the 5-year survival is estimated to be 33%. When this patient undergoes lobectomy for pathologic tumor stage IB, the 5-year survival is estimated to be 45% (Figure 2).

View larger version (20K): [in this window] [in a new window]   Figure 1. Preoperative mode of the prognostic model: predicted survival of a 64-year old man with chronic obstructive pulmonary disease (COPD), congestive heart failure, and clinical tumor stage IB.

View larger version (20K): [in this window] [in a new window]   Figure 2. Postoperative mode of the prognostic model: predicted survival of a 64-year old male with chronic obstructive pulmonary disease (COPD) and congestive heart failure who underwent lobectomy for pathologic tumor stage IB.

	Discussion

Top Abstract Introduction Methods Results Discussion References

 
In terms of survival, NSCLC is a heterogeneous disease with a remarkable variation in outcome among individual patients. During the past decades, researchers have attempted to identify factors that can be used to predict outcome of patients with NSCLC. Besides tumor stage, other factors, such as comorbidity and extent of resection, also have been proven to predict survival after NSCLC surgery.^3-6 In previously published studies we^8,9 found the CCI to predict postoperative outcome more accurately than the individual comorbid conditions. However, this comorbidity index was not developed specifically for patients having NSCLC surgery. Therefore, only individual comorbid conditions of the CCI that were associated with an impaired survival in the multivariate analysis were included in our prognostic model. In addition, the CCI does not include several other prognostic factors in NSCLC surgery patients, such as sex, age, extent of resection, and tumor stage, and it does not estimate the long-term survival of individual patients after surgical resection.

We developed a prognostic model with a preoperative and postoperative mode, including tumor-related factors, treatment-related factors, and clinical variables, for prediction of survival of individual patients after NSCLC surgery. We included prognostic factors that are all readily available and interpretable to the clinician. Most factors have been recognized as predictive in previous studies on patients operated on for NSCLC.^1,2,4,5 We note that if some apparently obvious risk factor does not appear significantly predictive (P < .05) in a multivariate model, such as congestive heart failure and COPD in our study, one cannot conclude that this is irrelevant to outcome and exclude this particular risk factor from the prognostic model.

The postoperative mode of the prognostic model, including pathologic tumor stage and type of resection, provided substantially better discrimination than a comorbidity index such as the CCI (c statistic 0.65 vs 0.55, P < .0001), the preoperative mode (c statistic 0.65 vs 0.61, P = .0002), or only pathologic tumor stage (c statistic 0.65 vs 0.60, P < .0001). Nevertheless, it is widely recognized that pathologic tumor stage is the most powerful predictor of long-term survival and the present postoperative mode of the prognostic model provides only slightly better discrimination than only pathologic tumor stage. Moreover, owing to similar relative risks of some tumor stages in our multivariate analysis, we combined these tumor stages in one subgroup. More patient data will be needed to improve the accuracy of this postoperative mode of the model.

Despite the postoperative mode being the most accurate in prediction of survival, it is important to select patients with a poor prognosis before surgery without knowledge of type of resection and pathologic tumor stage. For this purpose, the preoperative mode is more appropriate, which proved to be as accurate as only pathologic tumor stage in prediction of survival (c statistic 0.61 vs 0.60, P = .90) and more accurate than the CCI (c statistic 0.61 vs 0.55, P < .0001). Therefore, the preoperative mode can be used as a first evaluation to estimate survival of an individual patient and identify patients whose prognosis is poor. Also in this mode of the prognostic model, some tumor stages are combined in one subgroup and more patient data will be necessary to improve the accuracy of this preoperative model. The postoperative mode can either be used preoperatively to estimate survival assuming the type of resection planned and assuming that clinical tumor stage will not alter after resection, or it can be used postoperatively when type of resection and pathologic tumor stage are definite. A practical version of the present prognostic model for easy access and use by clinicians is available on the Internet (http://www.cardiothoracicresearch.nl).

Although bootstrapping techniques were used for internal validation of the model, a limitation of our study is that the prognostic model is not validated by an external test group. This may be essential before further clinical application is initiated.²⁰ In addition, this study is limited by the relatively small amount of clinical data that was pooled from only two centers, in terms of both number of data and availability of data. For example, histopathologic cell type failed to predict survival in our analysis, whereas others have shown an improved survival for squamous cell histology over adenocarcinoma, large cell carcinoma, and bronchoalveolar cell carcinoma.^10,21-23 Moreover, as we stated before, we combined some tumor stages in one subgroup. We plan to further improve and validate the prognostic model using data from other clinics.

In conclusion, we developed a simple prognostic model with a preoperative and postoperative mode that may be used for risk assessment in individual patients. The prognostic model successfully estimates long-term survival of individual patients and performed well with assessments of internal validity, with the postoperative mode being the most accurate. We therefore are confident that the model will also perform well for future patients who face the choice between operative treatment and other treatment modalities for NSCLC. Inclusion of more factors with additional prognostic value could potentially further improve the accuracy of the model. If further validated, this prognostic model could help clinicians and patients in clinical decision-making and treatment tailoring based on the estimated survival after surgery.

	Footnotes

 
Address for reprints: Erasmus MC Rotterdam, Dr. Molewaterplein 40, 3015 GD, Rotterdam, The Netherlands.

	References

Top Abstract Introduction Methods Results Discussion References

 

1. Mountain CF. Revisions in the International System for Staging Lung Cancer. Chest 1997;111:1710-1717.[Abstract/Free Full Text]
   
2. Lopez-Encuentra A, Garcia-Lujan R, Rivas JJ, Rodriguez-Rodriguez J, Torres-Lanza J, Varela-Simo G. Comparison between clinical and pathologic staging in 2,994 cases of lung cancer. Ann Thorac Surg 2005;79:974-979discussion 979.[Abstract/Free Full Text]
   
3. Battafarano RJ, Piccirillo JF, Meyers BF, Hsu HS, Guthrie TJ, Cooper JD, et al. Impact of comorbidity on survival after surgical resection in patients with stage I non–small cell lung cancer. J Thorac Cardiovasc Surg 2002;123:280-287.[Abstract/Free Full Text]
   
4. Sekine Y, Behnia M, Fujisawa T. Impact of COPD on pulmonary complications and on long-term survival of patients undergoing surgery for NSCLC. Lung Cancer 2002;37:95-101.[Medline]
   
5. Lopez-Encuentra A, Astudillo J, Cerezal J, Gonzalez-Aragoneses F, Novoa N, Sanchez-Palencia A. Prognostic value of chronic obstructive pulmonary disease in 2994 cases of lung cancer. Eur J Cardiothorac Surg 2005;27:8-13.[Abstract/Free Full Text]
   
6. van Meerbeeck JP, Damhuis RA, Vos de Wael ML. High postoperative risk after pneumonectomy in elderly patients with right-sided lung cancer. Eur Respir J 2002;19:141-145.[Abstract/Free Full Text]
   
7. Janssen-Heijnen ML, Schipper RM, Razenberg PP, Crommelin MA, Coebergh JW. Prevalence of co-morbidity in lung cancer patients and its relationship with treatment. a population-based study. Lung Cancer 1998;21:105-113.[Medline]
   
8. Birim O, Maat AP, Kappetein AP, van Meerbeeck JP, Damhuis RA, Bogers AJ. Validation of the Charlson comorbidity index in patients with operated primary non-small cell lung cancer. Eur J Cardiothorac Surg 2003;23:30-34.[Abstract/Free Full Text]
   
9. Birim O, Zuydendorp HM, Maat AP, Kappetein AP, Eijkemans MJ, Bogers AJ. Lung resection for non-small-cell lung cancer in patients older than 70. mortality, morbidity, and late survival compared with the general population. Ann Thorac Surg 2003;76:1796-1801.[Abstract/Free Full Text]
   
10. Firat S, Bousamra M, Gore E, Byhardt RW. Comorbidity and KPS are independent prognostic factors in stage I non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 2002;52:1047-1057.[Medline]
    
11. Ferguson MK, Durkin AE. A comparison of three scoring systems for predicting complications after major lung resection. Eur J Cardiothorac Surg 2003;23:35-42.[Abstract/Free Full Text]
    
12. Pauwels RA, Buist AS, Calverley PM, Jenkins CR, Hurd SS. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary. Am J Respir Crit Care Med 2001;163:1256-1276.[Free Full Text]
    
13. The World Health Organization histological typing of lung tumours. Second edition. Am J Clin Pathol. 1982;77:123-136.[Medline]
    
14. Steyerberg EW, Eijkemans MJ, Harrell Jr FE, Habbema JD. Prognostic modelling with logistic regression analysis. a comparison of selection and estimation methods in small data sets. Stat Med 2000;19:1059-1079.[Medline]
    
15. Ambrogi V, Pompeo E, Elia S, Pistolese GR, Mineo TC. The impact of cardiovascular comorbidity on the outcome of surgery for stage I and II non-small-cell lung cancer. Eur J Cardiothorac Surg 2003;23:811-817.[Abstract/Free Full Text]
    
16. Bernard A, Ferrand L, Hagry O, Benoit L, Cheynel N, Favre JP. Identification of prognostic factors determining risk groups for lung resection. Ann Thorac Surg 2000;70:1161-1167.[Abstract/Free Full Text]
    
17. Harrell Jr FE, Lee KL, Califf RM, Pryor DB, Rosati RA. Regression modelling strategies for improved prognostic prediction. Stat Med 1984;3:143-152.[Medline]
    
18. Steyerberg EW, Harrell Jr FE, Borsboom GJ, Eijkemans MJ, Vergouwe Y, Habbema JD. Internal validation of predictive models. efficiency of some procedures for logistic regression analysis. J Clin Epidemiol 2001;54:774-781.[Medline]
    
19. Harrell Jr FE, Lee KL, Mark DB. Multivariable prognostic models. issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361-387.[Medline]
    
20. Justice AC, Covinsky KE, Berlin JA. Assessing the generalizability of prognostic information. Ann Intern Med 1999;130:515-524.[Abstract/Free Full Text]
    
21. Mountain CF, Lukeman JM, Hammar SP, Chamberlain DW, Coulson WF, Page DL, Victor TA, Weiland LH. Lung cancer classification. the relationship of disease extent and cell type to survival in a clinical trials population. J Surg Oncol 1987;35:147-156.[Medline]
    
22. Padilla J, Calvo V, Penalver JC, Zarza AG, Pastor J, Blasco E, Paris F. Survival and risk model for stage IB non-small cell lung cancer. Lung Cancer 2002;36:43-48.[Medline]
    
23. Read RC, Yoder G, Schaeffer RC. Survival after conservative resection for T1 N0 M0 non-small cell lung cancer. Ann Thorac Surg. 1990;49:391-398discussion 399-400.[Abstract]
    

This article has been cited by other articles:

				M. F. Berry, J. Hanna, B. C. Tong, W. R. Burfeind Jr, D. H. Harpole, T. A. D'Amico, and M. W. Onaitis Risk factors for morbidity after lobectomy for lung cancer in elderly patients. Ann. Thorac. Surg., October 1, 2009; 88(4): 1093 - 1099. [Abstract] [Full Text] [PDF]

				A. Brunelli, A. Charloux, C. T. Bolliger, G. Rocco, J-P. Sculier, G. Varela, M. Licker, M. K. Ferguson, C. Faivre-Finn, R. M. Huber, et al. ERS/ESTS clinical guidelines on fitness for radical therapy in lung cancer patients (surgery and chemo-radiotherapy) Eur. Respir. J., July 1, 2009; 34(1): 17 - 41. [Abstract] [Full Text] [PDF]

				P. M. B. van Tilburg, H. Stam, H. C. Hoogsteden, and R. J. van Klaveren Pre-operative pulmonary evaluation of lung cancer patients: a review of the literature Eur. Respir. J., May 1, 2009; 33(5): 1206 - 1215. [Abstract] [Full Text] [PDF]

				D. C. Rice, C. W. Stevens, A. M. Correa, A. A. Vaporciyan, A. Tsao, K. M. Forster, G. L. Walsh, S. G. Swisher, W. L. Hofstetter, R. J. Mehran, et al. Outcomes After Extrapleural Pneumonectomy and Intensity-Modulated Radiation Therapy for Malignant Pleural Mesothelioma Ann. Thorac. Surg., November 1, 2007; 84(5): 1685 - 1693. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to Personal Folders Download to citation manager Author home page(s): A. Pieter Kappetein Michel I.M. Versteegh Ad J.J.C. Bogers Permission Requests Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Birim, O. Articles by Bogers, A. J.J.C. Search for Related Content PubMed PubMed Citation Articles by Birim, O. Articles by Bogers, A. J.J.C. Related Collections Lung - cancer