The Effect of Tumor Size on Curability of Stage I Non-small Cell Lung Cancers* Juan P. Wisnivesky, MD, MPH; David Yankelevitz, MD; and Claudia I. Henschke, PhD, MD, FCCP
Objective: The objective of this study was to determine the relationship between tumor size and curability of stage I non-small cell lung cancer. Methods: From the Surveillance, Epidemiology, and End Results registry 2003, we identified all primary non-small cell lung cancer cases that were diagnosed prior to autopsy. Among these cases, we narrowed the focus to those diagnosed in 1988 or later, and to 7,620 patients who had undergone curative surgical resection. Kaplan-Meier survival curves were obtained for these stage I malignancies for five tumor size categories (ie, 5 to 15 mm, 16 to 25 mm, 26 to 35 mm, 36 to 45 mm, and > 45 mm). The 12-year Kaplan-Meier estimator of survival was used as a measure of lung cancer cure rate. Results: Among 7,620 stage I cancers, cure rates decreased with increasing tumor size. The 12-year survival rates for patients with tumors 5 to 15 mm in diameter was 69% (95% confidence interval [CI], 64 to 74%), 63% for those with tumors 16 to 25 mm in diameter (95% CI, 60 to 67%), 58% for those with tumors 26 to 35 mm in diameter (95% CI, 54 to 61%), 53% for those with tumors 36 to 45 mm in diameter (95% CI, 48 to 57%), and 43% for those with tumors > 45 mm in diameter (95% CI, 39 to 48%). Cure rates were statistically significantly different for all tumor size categories (p < 0.05) except for the groups with tumors 26 to 35 mm and 36 to 45 mm in diameter (p ⴝ 0.10). Conclusions: Smaller tumor size at diagnosis is associated with improved curability within stage I non-small cell lung cancers. These results suggest that further subclassification by size within stage I may be important. (CHEST 2004; 126:761–765) Key words: curability; non-small cell lung cancer; stage I; survival Abbreviations: CI ⫽ confidence interval; SEER ⫽ Surveillance, Epidemiology, and End Results
lung cancer cure rate is dismal and has T heremained essentially unchanged over the past 3 decades.1 The stage of disease at diagnosis represents one of the most powerful determinants of outcome in patients with non-small cell lung cancer, with patients having earlier stage disease having a better chance of long-term survival.2 Unfortunately,
*From the Division of General Internal Medicine and Pulmonary, Critical Care, and Sleep Medicine (Dr. Wisnivesky), Mount Sinai School of Medicine, and the Department of Radiology (Drs. Yanelevitz and Henschke), New York-Presbyterian HospitalWeill Cornell Medical Center, New York, NY. Manuscript received December 16, 2003; revision accepted April 21, 2004. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail:
[email protected]). Correspondence to: Juan P. Wisnivesky, MD, MPH, Department of Medicine, Mount Sinai School of Medicine, One Gustave L. LevyPlace,Box1087,NewYork,NY10029;e-mail:juan.wisnivesky@ mssm.edu www.chestjournal.org
only small proportions of non-small cell lung cancer cases are diagnosed at an earlier stage. The importance of tumor size on the survival of patients with early non-small cell lung cancers was recognized in the 1997 American Joint Committee on Cancer staging system,2 as stage I was subdivided into stage IA (tumors ⬍ 3 cm in diameter) and stage IB (tumors ⬎ 3 cm in diameter). However, controversy remains concerning the relationships between tumor size and prognosis, and the appropriate cutoff size at which to stratify patients within stage I.3 Although some studies4,5 have shown better survival for patients with stage I tumors measuring ⬍ 2 cm, or even 1 cm, others have not.3,6 Smaller tumor size at the time of diagnosis was found to be associated with an improved 5-year survival rate in some of these studies. However, the 5-year survival rate of patients with smaller tumors may be higher as a consequence of lead time even if early diagnosis does not increase the chance of cure.7 Thus, the relationCHEST / 126 / 3 / SEPTEMBER, 2004
761
ship of tumor size and curability cannot be inferred from the results of these studies. This study was conducted to better determine the effect of tumor size on the curability of stage I cancers.
Materials and Methods The Surveillance, Epidemiology, and End Results (SEER) program is a National Cancer Institute-funded database that has been collecting clinicopathologic data on all incident cancer cases in selected geographic areas of the United States since 1973. From the SEER 9 registry 2003, we identified all cases of non-small cell lung cancer (tumor site codes, 34.0 to 34.9; International Classification of Diseases for Oncology-2 morphology codes, 8010 to 8040, 8050 to 8076, 8140, 8143, 8250 to 8260, 8310, 8320, 8323, 8470 to 8490, and 8550 to 8573) that were diagnosed prior to autopsy.8 Among these cases, we narrowed the focus to those diagnosed in 1988 or later, as the histologic classification had changed. There were 143,796 such cases. Among these 143,796 cases, we identified cases of primary stage I cancer using the American Joint Committee on Cancer criteria,2 as follows: T1, noninvasive tumor with a diameter of ⱕ 3 cm, without evidence of invasion more proximal than the lobar bronchus (SEER tumor extent code 10); T2, tumor with any of the following features among ⬎ 3 cm in dimension, involvement of main bronchus ⱖ 2 cm distal to the carina, invasion of the visceral pleura, association with atelectasis or obstructive pneumonitis that extends to the hilar region but does not involve the entire lung (SEER tumor extent codes 20 and 40); N0, absence of lymph node metastasis (SEER lymph node code 0); and M0, no distant metastasis (SEER tumor extension ⬍ 40). We found 11,142 cases of T1N0M0 or T2N0M0 cancer. Among these stage I cases, 7,620 patients underwent resection (ie, lobectomy at least; site-specific surgery codes 30 to 70). Histology was identified according to the World Health Organization definition published in 1990.9 Tumor size, as defined by its greatest diameter, was based on those recorded in the pathology report. Cases were classified into the following five tumor size (ie, diameter) categories: 5 to 15 mm; 16 to 25 mm; 26 to 35 mm; 36 to 45 mm; and ⬎ 45 mm. We observed a round-off pattern with respect to the reporting of tumor size in the SEER database. Thus, size categories were defined so that tumors that were clustered together due to this phenomenon would be placed in the same group. We constructed lung-cancer specific survival curves separately by tumor size category using the Kaplan-Meier method, which adjusted for deaths from all other causes.10 The cause of death provided in the SEER registry is abstracted from the National Center for Health Statistics database of consolidated death certificates from the vital statistics office in each state. As previously defined by Buell,11 the cure rate can be determined by the point at which the cumulative survival no longer changes, that is, when the slope of the survival curve approaches zero. Based on this definition, we used the 12-year cumulative lung cancer survival rate as a measure of the cure rate, given that the slope of the survival curves for all tumor size categories approached zero before this time point. Differences in 12-year survival rates were evaluated using a z-test based on the large sample distribution of the Kaplan-Meier estimators.12 The analyses were performed using a statistical software package (SAS; SAS Institute; Cary, NC). 762
Results Of the 7,620 patients with stage I non-small cell lung cancer, 3,273 (43%) were female. The mean (⫾ SD) age was 66 ⫾ 9 years. The histologic classification of the tumors showed that there were 2,611 cases of squamous cell carcinoma (34%), 3,319 cases of adenocarcinoma (44%), and 516 cases of large cell carcinoma (7%). Among the 7,620 cases, we found 786 (10%) to be 5 to 15 mm in diameter, 2,290 (30%) to be 16 to 25 mm in diameter, 2,006 (26%) to be 26 to 35 mm in diameter, 1,100 (15%) to be 36 to 45 mm in diameter, and 1,438 (19%) to be ⬎ 45 mm in diameter. The Kaplan-Meier survival curves stratified by tumor size category are shown in Figure 1. Among these 7,620 stage I cancers, cure rates decreased with increasing tumor size. The lung cancer cure rates were as follows: tumors 5 to 15 mm in diameter, 69% (95% confidence interval [CI], 64 to 74%); tumors 16 to 25 mm in diameter, 63% (95% CI, 60 to 67%); tumors 26 to 35 mm in diameter, 58% (95% CI, 54 to 61%); tumors 36 to 45 mm in diameter, 53% (95% CI, 48 to 57%); and tumors ⬎ 45 mm in diameter, 43% (95% CI, 39 to 48%) [Table 1]. A comparison of cure rates according to tumor size category showed that rates were statistically significantly different for all categories (p ⬍ 0.05), except for tumor sizes 26 to 35 mm and 36 to 45 mm (p ⫽ 0.10). Discussion While differences in survival between patients with tumors ⬍ 3 cm and ⬎ 3 cm in diameter have been determined in prior studies, a similar relationship for lesions of different sizes within stage I, particularly tumors ⬍ 3 cm in diameter, has not been established. The results of this study show that smaller tumor size at diagnosis is associated with improved curability for patients with stage I nonsmall cell lung cancers. Disease-specific cure rates for lung cancer ranged from as high as 69% (95% CI, 64 to 74%) for malignancies 5 to 15 mm in diameter to approximately 43% (95% CI, 39 to 48%) for those ⬎ 45 mm in diameter. The observed association between smaller tumor size and improved survival can be explained by several possible mechanisms. First, if a lung cancer is diagnosed when it is small instead of when it is large, then survival, which is measured from the time of diagnosis, should be increased by the lead time from earlier diagnosis.7 However, as shown in Figure 1, the slope of the survival curves for all tumor size categories approached zero, suggesting that by 12 years after diagnosis even the smallest tumors that Clinical Investigations
Figure 1. Cure rates according to tumor size category were statistically significantly different for all categories (p ⬍ 0.05) except for groups of patients with tumor diameters of 26 to 35 mm and 36 to 45 mm (p ⫽ 0.10).
had metastasized before resection had sufficient time to progress and cause the death of the individual. The survival curves for all tumor sizes become almost parallel beyond this time point, suggesting that differences in cumulative survival represent true differences in long-term outcomes. Thus, the 12year survival rate, our measure of curability, should have not been subject to lead-time bias. Second, it has been suggested that the smaller the tumor size at detection the greater the probability that the lesion is a case of overdiagnosis. These overdiagnosed lesions are defined as tumors that, although pathologically classified as cancer, would not result in the death of the person.13,14 Because overdiagnosed cases will effectively dilute lung cancer cases with healthy individuals, it can artificially increase the observed survival of the smaller tumors. We have previously shown using the same cancer registry that the casefatality rate of untreated stage I lung cancer appears to be of the order of 90% even when the tumor diameter is ⬍ 15 mm.15 Thus, our finding of im-
Table 1—Cure Rates for Non-small Cell Lung Cancers According to Tumor Size Tumor Size, mm 5–15 16–25 26–35 36–45 ⬎45
www.chestjournal.org
Patients, No. (%)
Cure Rate, % (95% CI)
786 (10) 2,290 (30) 2,006 (26) 1,100 (15) 1,438 (19)
69 (64–74) 63 (60–67) 58 (54–61) 53 (48–57) 43 (39–48)
proved curability with smaller tumor size is unlikely to be explained by the possibility of overdiagnosis. The most probable interpretation of our results is that earlier treatment increases the probability of curing malignant and potentially fatal lung cancer. The prognostic value of tumor size within stage I has been evaluated in several studies.3,4,16 –21 In a study including 286 stage I tumors, Watanabe et al16 showed significant differences in 5-year survival rates for patients with tumors ⬍ 3 cm, 3 to 5 cm, and ⬎ 5 cm in diameter. Similarly, Carbone et al17 showed that patients with stage I cancers ⬎ 5 cm in diameter had a significantly poorer prognosis and proposed to change the classification of these lesions. Additionally, several studies have shown that there is important variation in the 5-year survival rates of stage IA subgroups. Recently, Gajra et al4 found in a series of 246 stage IA cases that patients with tumors ⬍ 1.5 cm in diameter had an improved outcome compare to those with tumors ⬎ 1.5 cm in size. Koike et al5 found a significant difference in survival between patients with non-small cell lung cancers ⬍ 2 cm in diameter and those with lung cancers ⬎ 2 cm in diameter, but tumor size was not a significant prognostic factor on multivariate analysis. Similarly, Ishida et al,19 Read et al,18 Riquet et al,20 and Padilla et al21 found significant differences in 5-year survival rates between patients with tumors ⬍ 2 cm in diameter and those with tumors measuring 2.1 to 3 cm. Harpole et al22 compared the 5-year survival rates of groups with tumor sizes of ⬍ 2 cm, 2 to 4 cm, and ⱖ 4 cm. There were significant differences between CHEST / 126 / 3 / SEPTEMBER, 2004
763
each group. However, as some of these studies utilized 5-year survival rates as the measure of prognosis and used the log rank test or Cox regression to compare different tumor size categories, their results may have been confounded by lead-time bias. Conversely, other studies have found no difference in the survival of stage I lung cancer patients according to tumor size. Yanagi et al6 found no significant difference in survival between patients with lesions ⬍ 2 cm in diameter and those with lesions ⬎ 2 cm in diameter among 64 cases of non-small cell lung cancer. Similarly, Patz et al,3 using a Cox proportional hazard model, found no survival difference according to tumor size among 510 patients with pathologic stage IA non-small cell lung cancers, of which only a minority (approximately 10%) were ⬍ 1 cm in diameter. Thus, the study may not have had sufficient power to detect meaningful differences in survival.23 Our results suggest that further subclassification by tumor size within stage I may be important, as already has been done in the present TNM classification of the tumors of oral cavity, pharynx, thyroid gland, and breast.24 Based on the growing evidence that tumor size is a key prognostic indicator of lung cancer survival, accurate imaging and pathologic measurements should be obtained in patients with these malignancies. Ideally, the length, width, and height of the tumor should be measured according to a standardized, well-defined measurement approach. Some studies25–30 using low-dose CT scan screening showed that low-dose CT scanning is substantially more sensitive than a chest radiograph for the detection of early lung cancer. The majority of the lung cancers detected with CT scanning in these studies were of the earliest stage (ie, stage IA) and typically were ⬍ 10 mm in diameter, a marked contrast to the distribution of lung cancers detected in clinical practice. For lung cancer screening to be effective, however, the ability of CT scans to detect smaller lesions has to translate into a decrease in lung cancer deaths or an increase in its complement, the cure rate. The results of this study suggest that early detection increases the curability of clinically diagnosed lung cancer. Applying these results to cases detected by screening, however, requires consideration of other issues. First, this study was limited to surgically resected stage I cases. In a screening program, some patients may refuse or be ineligible for surgery, and, thus, the tumor sizespecific cure rates for screening-detected cases may be lower than those estimated in the study. Conversely, because, overall, tumors in screeningdetected cases will tend to be smaller (mostly ⬍ 10 mm) and the patients will be asymptomatic, the opposite phenomenon may be observed. Second, to 764
evaluate the effectiveness of screening, the effect of competing causes of death on the overall survival rate needs to be taken into consideration. Information regarding the cause of death for patients in the SEER registry is abstracted from death certificates. Although the death certificate is an important source of data on disease incidence, prevalence, and mortality, inaccuracies in the reported cause of death in this document have been described.31,32 However, our estimates of lung cancer survival rates using the SEER database are similar to those reported in other studies using hospital-based cases series.2 Moreover, the large number of patients with stage I lung cancer in SEER database allow for a precise estimation of cure rate in these patients. In summary, tumor size at diagnosis among patients with stage I lung cancer is associated with the probability of cure. These results suggests that further subclassification by size within stage I may be important. References 1 Smith RA, Glynn TJ. Epidemiology of lung cancer. Radiol Clin North Am 2000; 38:453– 470 2 Mountain CF. Revisions in the international system for staging lung cancer. Chest 1997; 111:1710 –1717 3 Patz EF Jr, Rossi S, Harpole DH Jr, et al. Correlation of tumor size and survival in patients with stage IA non-small cell lung cancer. Chest 2000; 117:1568 –1571 4 Gajra A, Newman N, Gamble GP, et al. Impact of tumor size on survival in stage IA non-small cell lung cancer: a case for subdividing stage IA disease. Lung Cancer 2003; 42:51–57 5 Koike T, Terashima M, Takizawa T, et al. Clinical analysis of small-sized peripheral lung cancer. J Thorac Cardiovasc Surg 1998; 115:1015–1020 6 Yanagi S, Sugiura H, Morikawa T, et al. Tumor size does not have prognostic significance in stage Ia NSCLC. Anticancer Res 2000; 20:1155–1158 7 Morrison A, Screening. Rothman KJ, et al, eds. Modern epidemiology. Philadelphia, PA: Lippincott-Raven, 1998 8 Ries LAG, Eisner MP, Kosary CL, et al, eds. SEER Cancer Statistics Review, 1975–2001; Bethesda, MD: National Cancer Institute. Available at http://seer.cancer.gov/csr/1975– 2001/. Accessed August 11, 2004. 9 Percy C VHV, Muir C, eds. International classification of diseases for oncology. 2nd ed. Geneva, Switzerland: World Health Organization, 1990 10 Kaplan EL, Meier P. Nonparametric estimation for incomplete obserbations. J Am Stat Assoc 1958; 53:457– 481 11 Buell PE. The importance of tumor size in prognosis for resected bronchogenic carcinoma. J Surg Oncol 1971; 3:539 – 551 12 Klein J, Moeschberger ML. Survival analysis: techniques for censored and truncated data. New York, NY: Springer-Verlag, 1997 13 Bailar JC III. Screening for lung cancer: where are we now? Am Rev Respir Dis 1984; 130:541–542 14 Black WC, Welch HG. Advances in diagnostic imaging and overestimations of disease prevalence and the benefits of therapy. N Engl J Med 1993; 328:1237–1243 Clinical Investigations
15 Henschke CI, Wisnivesky JP, Yankelevitz DF, et al. Small stage I cancers of the lung: genuineness and curability. Lung Cancer 2003; 39:327–330 16 Watanabe Y, Shimizu J, Oda M, et al. Proposals regarding some deficiencies in the new international staging system for non-small cell lung cancer. Jpn J Clin Oncol 1991; 21:160 – 168 17 Carbone E, Asamura H, Takei H, et al. T2 tumors larger than five centimeters in diameter can be upgraded to T3 in non-small cell lung cancer. J Thorac Cardiovasc Surg 2001; 122:907–912 18 Read RC, Yoder G, Schaeffer RC. Survival after conservative resection for T1 N0 M0 non-small cell lung cancer. Ann Thorac Surg 1990; 49:391–398 19 Ishida T, Yano T, Maeda K, et al. Strategy for lymphadenectomy in lung cancer three centimeters or less in diameter. Ann Thorac Surg 1990; 50:708 –713 20 Riquet M, Manac’h D, Le Pimpec Barthes F, et al. Prognostic value of T and N in non small cell lung cancer three centimeters or less in diameter. Eur J Cardiothorac Surg 1997; 11:440 – 443 21 Padilla J, Calvo V, Penalver JC, et al. Surgical results and prognostic factors in early non-small cell lung cancer. Ann Thorac Surg 1997; 63:324 –326 22 Harpole DH Jr, Herndon JE II, Young WG Jr, et al. Stage I nonsmall cell lung cancer: a multivariate analysis of treatment methods and patterns of recurrence. Cancer 1995; 76:787– 796 23 Black WC. Unexpected observations on tumor size and
www.chestjournal.org
24 25
26
27 28
29
30
31 32
survival in stage IA non-small cell lung cancer. Chest 2000; 117:1532–1534 Watanabe Y. Substaging of stage I: a commentary. Lung Cancer 2003; 42:59 – 61 Henschke CI, McCauley DI, Yankelevitz DF, et al. Early Lung Cancer Action Project: overall design and findings from baseline screening. Lancet 1999; 354:99 –105 Henschke CI, Naidich DP, Yankelevitz DF, et al. Early lung cancer action project: initial findings on repeat screenings. Cancer 2001; 92:153–159 Swensen SJ. CT screening for lung cancer. AJR Am J Roentgenol 2002; 179:833– 836 Diederich S, Wormanns D, Semik M, et al. Screening for early lung cancer with low-dose spiral CT: prevalence in 817 asymptomatic smokers. Radiology 2002; 222:773–781 Sone S, Li F, Yang ZG, et al. Results of three-year mass screening programme for lung cancer using mobile low-dose spiral computed tomography scanner. Br J Cancer 2001; 84:25–32 Sobue T, Moriyama N, Kaneko M, et al. Screening for lung cancer with low-dose helical computed tomography: anti-lung cancer association project. J Clin Oncol 2002; 20:911–920 Smith Sehdev AE, Hutchins GM. Problems with proper completion and accuracy of the cause-of-death statement. Arch Intern Med 2001; 161:277–284 Kircher T, Nelson J, Burdo H. The autopsy as a measure of accuracy of the death certificate. N Engl J Med 1985; 313:1263–1269
CHEST / 126 / 3 / SEPTEMBER, 2004
765