- Email: [email protected]
Uneven Distribution of Cancer Histology in the National Lung Screening Trial
REVIEW ARTICLE
Uneven Distribution of Cancer Histology in the National Lung Screening Trial Thomas A. Dillard, MD, Raxit R. Patel, MD and Carsten Schroeder, MD
Abstract: The National Lung Screening Trial reported a 20% lower mortality due to lung cancer in the patients screened with low-dose computed tomography (LDCT) compared with plain chest radiography (XRAY). A hypothesis was raised that LDCT should detect more cases of all tissue types or else the distribution of tissue types should be equal between groups. Data were extracted regarding the tissue types of lung cancer and presenting stages from the 2011 NSLT report. A total of 1,993 cases of tissue diagnosed lung cancer were reported, 1,054 for the LDCT group and 939 for the XRAY group. Two tissue types were more prevalent in the XRAY group: small cell carcinomas (16.9% versus 13%; P , 0.05) and other non-small cell (ONSC) carcinomas (16.8% versus 12.4%; P , 0.05). The ONSC category excluded the usual nonsmall cell tissue types: adenocarcinomas, squamous, and large cell carcinomas and did include other known biologically unfavorable tissue types. The XRAY group also had a disproportionately large number of stage IV small cell and ONSC tumors (P , 0.05 for ONSC). Bronchoalveolar cell carcinomas were more prevalent in the LDCT group (10.4% versus 3.7%, P , 0.05), likely reflecting greater sensitivity for detection. In summary, this review found uneven distribution of cases and higher preponderance of stage IV tumors for 2 adverse tissue types in the XRAY group. The results are consistent with greater severity of disease in the XRAY group with potential for length time bias and reduced mortality benefit from LDCT screening. Key Indexing Terms: Lung cancer; Low-dose CT scanning; Chest x-ray; Cancer screening. [Am J Med Sci 2015;350(3):219–221.]
I
n December 2013, the U.S. Preventive Services Task Force (USPSTF) recommended annual screening for lung cancer with low-dose computed tomography (LDCT) in adults aged 55 to 80 years who have a 30 pack-year smoking history and currently smoke or have quit within the past 15 years.1 The USPSTF decision was based primarily on the National Lung Screening Trial (NLST),2 which reported a 20% lower mortality due to lung cancer in the patients screened with LDCT compared with plain chest radiography (XRAY). The NLST is the largest study to report a positive finding in support of screening for lung cancer in contrast to earlier studies in the United States and abroad.3 The NLST study performed lung cancer screening for more than 53,000 active or former smokers with a 30 pack-year or more cigarette smoking exposure and age between 55 and 74 years. Study subjects were randomized to receive LDCT scans or plain chest radiographs at entry (T0) and annually for 2 years (T1 and T2). Subjects received clinical follow-up annually for a total of 5 to 7 years as warranted clinically but without further From the Georgia Regents University (TAD, RRP, CS), Augusta, Georgia; Pulmonary, Critical Care and Sleep Division (TAD, RRP), Department of Medicine; and Department of Surgery (CS), Georgia Regents University Cancer Center, Augusta, Georgia. Submitted April 13, 2015; accepted in revised form May 8, 2015. The authors have no other conflicts of interest to disclose. Supported by Georgia Regents University, Augusta, GA. Correspondence: Thomas A. Dillard, MD, Georgia Regents University, 1120 15th Street, BBR-5513, Augusta, GA 30912 (E-mail: [email protected]).
The American Journal of the Medical Sciences
imaging within the study. In most cases, the tissue diagnosis of cancer was made 3 to 5 years after NLST study screening phase ended. Actions such as diagnostic imaging or biopsy followed by usual treatment were taken as indicated by the imaging findings during the study. The 2011 report cited 356 deaths in the LDCT group and 443 deaths in the XRAY group attributable to lung cancer as determined by the end point verification team, a difference of 87 fewer deaths. The difference was 76 in the death certificate analysis. Other causes of death, such as other cancer, cardiovascular disease, respiratory illness and others, accounted for the majority of deaths. The total number of deaths by all causes was reported as 3,856. The NLST included the histologic types of cancers diagnosed during the course of study after randomization to either LDCT or XRAY screening groups. Obviously, tissue type could not be a condition for stratification at randomization before screening; however, histologic type could have a major bearing on outcomes such as mortality within the studied time frame. Depending on stage of disease, patients with squamous cell carcinoma or adenocarcinoma, for example, would typically survive longer than patients with small cell carcinoma and certain other less common histologic types of non-small cell (NSC) lung cancer cited below. This study aimed to compare the proportion of tumor histologic types in the NLST between the LDCT and XRAY screened populations. Given the greater sensitivity of LDCT scanning compared with XRAY, the hypothesis was raised that all histologic types of lung cancer would be detected more often in the LDCT group. The null hypothesis held that the proportion of each tissue type should be equal in the 2 screened groups.
METHODS Data were extracted regarding the tissue types of lung cancer diagnoses from the data published in the NLST report, in particular tables 5, 6 and 7 in that report. Data were excluded from consideration for 8 carcinoid cases in this report, 6 in the LDCT group and 2 in the XRAY group, due to the usual slow growth or benign nature of those tumors. Institutional review board approval was not required for this report. The NLST reported data for tissue types of lung cancers including bronchoalveolar cell carcinomas, adenocarcinomas, squamous cell carcinomas, large cell carcinomas, small cell carcinomas and other non-small cell carcinomas (ONSC). Clinicians have long grouped adenocarcinomas, squamous cell, bronchoalveolar cell and large cell carcinomas together as NSC carcinomas. The ONSC category in the NLST did not include the customary NSC types as noted above. Although the term bronchoalveolar cell carcinoma has been replaced in current nomenclature by the term adenocarcinoma in situ, the NLST considered each of these NSC tissue types separately as has this current report. In the NLST, ONSC carcinomas included the tissue types adenosquamous carcinomas (N 5 28), sarcomatoid carcinomas (N 5 6), unclassified carcinomas (N 5 55), anaplastic
Volume 350, Number 3, September 2015
219
Dillard et al
carcinomas (N 5 1), carcinosarcoma (N 5 1) and those coded only as non-small cell carcinoma (N 5 198). The ONSC tumor breakdown was not reported for each group separately. Many of these tumors have a known poor prognosis.4–10 Data analysis used 2 by 2 contingency tables with x2 statistics and Yates’ correction for continuity (Table 1). Fisher’s exact tests were used where appropriate. The authors report P values less than 0.05 as statistically significant.
RESULTS A total of 1,993 cases of tissue diagnosis of lung cancer were reported in the NLST, 1,054 for the LDCT group and 939 for the XRAY group. Table 1 shows the number of cases of each histologic type in the LDCT and XRAY screened groups of the NLST. The number of small cell carcinomas and ONSC tissue types in the XRAY group was higher than the LDCT screened group (Table 1, Figure 1), and the proportion was statistically significantly greater (P , 0.05). The LDCT group had a significantly larger proportion of bronchoalveolar cell carcinoma (P , 0.05). This is consistent with the greater sensitivity of LDCT for this tissue type. The proportions of adenocarcinomas, squamous cell and large cell carcinomas did not differ statistically significantly between groups (Table 1). Next, the distribution of cancer stages for the small cell and ONSC tumor types was examined. The proportion of stage IV tumors for these 2 tissue types combined was significantly greater in the XRAY (43 more cases) than the LDCT group (P , 0.05). Considered separately, the amount was significant for ONSC (P , 0.05; Figure 2).
DISCUSSION This focused review found a higher proportion of 2 adverse histologic types in the XRAY group of the NLST, small cell and ONSC carcinomas. In addition, a higher proportion of stage IV tumors with these tissue types was observed in the XRAY group. The disproportion of stage IV small cell and ONSC tumors in the XRAY group tumors also supports the contention of adverse biologic nature of these tissue types. These results run counter to the expectation of greater sensitivity with LDCT scanning. These also run counter to the null hypothesis that the default distribution of tissue types
FIGURE 1. Histologic types of lung cancer by screened group in the NLST. The proportion of BAC cases was significantly greater in the LDCT group. The proportion of ONSC and small cell cases in the XRAY group was statistically significantly higher. The other tissue types did not reach statistical significance. BAC, bronchoalveolar cell carcinoma; LDCT, low-dose computed tomography; NLST, National Lung Screening Trial; ONSC, other non-small cell.
would be equal in both groups. As stated above, tissue type could not be a condition for randomization before screening; however, the uneven results of tissue types have some implications on the apparent results of screening. One implication is the potential for length time bias in the NLST.11 Length time bias occurs when the disease severity is worse for one group compared with another. Aggressive tumors have a shorter natural history, leading to earlier death, than less aggressive tumors. Higher mortality in the XRAY group due to greater severity of disease would falsley increase the apparent mortality benefit of LDCT scanning. This effect confounds the interpretation of the lower mortality in the LDCT group as solely due to superior detection of lung cancers and earlier intervention. The foregoing discussion suggests that there may be less mortality benefit from LDCT screening than first reported; however, this current review considers only a single aspect of
TABLE 1. Histology of lung cancer diagnoses in the National Lung Screening Trial by screened group LDCT XRAY All Carcinomas BAC Adeno Squamous Large ONSC Small cell Unknown All
N
%
N
110 380 243 41 131 137 12 1,054
10.4 36.1 23.1 3.9 12.4 13 1.1 100
35 328 206 43 158 159 10 939
%
N
3.7 145 34.9 708 21.9 449 4.6 84 16.8 289 16.9 296 1.1 22 100 1,993
%
P
7.3 35.5 22.5 4.2 14.5 14.9 1.1 100
0.0001* 0.6064; NS 0.5552; NS 0.5031; NS 0.0061* 0.0139* 1.0000; NS —
Data for histology was taken from table 6 of the NLST. BAC, bronchoalveolar cell carcinoma; ONSC, other non-small cell carcinoma. * indicates P , 0.05
220
FIGURE 2. Stage at presentation of ONSC tumors in the NLST by screened group in the NLST. The XRAY group had significantly more stage IV ONSC tumors. The LDCT group had significantly more stage 1A tumors. Other stages did not reach statistical significance. LDCT, low-dose computed tomography; NLST, National Lung Screening Trial; ONSC, other non-small cell. Volume 350, Number 3, September 2015
Uneven Distribution of Cancer Histology
the NLST data and makes no attempt to refute a net benefit of screening. One purpose of this report is to elicit further discussion on the uneven distribution. In summary, this review performed a focused analysis of the distribution of cancer tissue types in the NLST. The main significance of this report is to call attention to the uneven distribution of tissue types, which have worse prognosis. ACKNOWLEDGMENTS The authors thank the National Cancer Institute for access to NCI’s data collected by the National Lung Screening Trial. The statements contained herein are solely those of the authors and do not represent or imply concurrence or endorsement by NCI. REFERENCES 1. Available at: http://www.uspreventiveservicestaskforce.org/uspstf/ uspslung.htm. Accessed January 8, 2014. 2. National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395–409. 3. Bach PB, Mirkin JN, Oliver TK, et al. Benefits and harms of CT screening for lung cancer: a systemic review. JAMA 2012;307: 2418–29.
Copyright © 2015 by the Southern Society for Clinical Investigation.
4. Available at: http://seer.cancer.gov/csr/1975_2010/browse_csr.php? sectionSEL515&pageSEL5sect_15_table.13.html. Accessed January 10, 2014. 5. Lee Y, Chung JH, Kim SE, et al. Adenosquamous carcinoma of the lung: CT, FDG PET, and clinicopathologic findings. Clin Nucl Med 2014;39:107–12. 6. Sökücü SN, Kocatürk C, Urer N, et al. Evaluation of six patients with pulmonary carcinosarcoma with a literature review. Scientific WorldJournal 2012;2012:167317:1–5. 7. Filosso PL, Ruffini E, Asioli S, et al. Adenosquamous lung carcinomas: a histologic subtype with poor prognosis. Lung Cancer 2011;74: 25–9. 8. Yendamuri S, Caty L, Pine M, et al. Outcomes of sarcomatoid carcinoma of the lung: a surveillance, epidemiology, and end results database analysis. Surgery 2012;152:397–402. 9. Venissac N, Pop D, Lassalle S, et al. Sarcomatoid lung cancer (spindle/giant cells): an aggressive disease? J Thorac Cardiovasc Surg 2007;134:619–23. 10. Martin LW, Correa AM, Ordonez NG, et al. Sarcomatoid carcinoma of the lung: a predictor of poor prognosis. Ann Thorac Surg 2007;84: 973–80. 11. Available at http://www.acponline.org/clinical_information/journals_ publications/ecp/marapr99/primer.htm Accessed January 15, 2014.
221
Uneven Distribution of Cancer Histology in the National Lung Screening Trial Thomas A. Dillard, MD, Raxit R. Patel, MD and Carsten Schroeder, MD
Abstract: The National Lung Screening Trial reported a 20% lower mortality due to lung cancer in the patients screened with low-dose computed tomography (LDCT) compared with plain chest radiography (XRAY). A hypothesis was raised that LDCT should detect more cases of all tissue types or else the distribution of tissue types should be equal between groups. Data were extracted regarding the tissue types of lung cancer and presenting stages from the 2011 NSLT report. A total of 1,993 cases of tissue diagnosed lung cancer were reported, 1,054 for the LDCT group and 939 for the XRAY group. Two tissue types were more prevalent in the XRAY group: small cell carcinomas (16.9% versus 13%; P , 0.05) and other non-small cell (ONSC) carcinomas (16.8% versus 12.4%; P , 0.05). The ONSC category excluded the usual nonsmall cell tissue types: adenocarcinomas, squamous, and large cell carcinomas and did include other known biologically unfavorable tissue types. The XRAY group also had a disproportionately large number of stage IV small cell and ONSC tumors (P , 0.05 for ONSC). Bronchoalveolar cell carcinomas were more prevalent in the LDCT group (10.4% versus 3.7%, P , 0.05), likely reflecting greater sensitivity for detection. In summary, this review found uneven distribution of cases and higher preponderance of stage IV tumors for 2 adverse tissue types in the XRAY group. The results are consistent with greater severity of disease in the XRAY group with potential for length time bias and reduced mortality benefit from LDCT screening. Key Indexing Terms: Lung cancer; Low-dose CT scanning; Chest x-ray; Cancer screening. [Am J Med Sci 2015;350(3):219–221.]
I
n December 2013, the U.S. Preventive Services Task Force (USPSTF) recommended annual screening for lung cancer with low-dose computed tomography (LDCT) in adults aged 55 to 80 years who have a 30 pack-year smoking history and currently smoke or have quit within the past 15 years.1 The USPSTF decision was based primarily on the National Lung Screening Trial (NLST),2 which reported a 20% lower mortality due to lung cancer in the patients screened with LDCT compared with plain chest radiography (XRAY). The NLST is the largest study to report a positive finding in support of screening for lung cancer in contrast to earlier studies in the United States and abroad.3 The NLST study performed lung cancer screening for more than 53,000 active or former smokers with a 30 pack-year or more cigarette smoking exposure and age between 55 and 74 years. Study subjects were randomized to receive LDCT scans or plain chest radiographs at entry (T0) and annually for 2 years (T1 and T2). Subjects received clinical follow-up annually for a total of 5 to 7 years as warranted clinically but without further From the Georgia Regents University (TAD, RRP, CS), Augusta, Georgia; Pulmonary, Critical Care and Sleep Division (TAD, RRP), Department of Medicine; and Department of Surgery (CS), Georgia Regents University Cancer Center, Augusta, Georgia. Submitted April 13, 2015; accepted in revised form May 8, 2015. The authors have no other conflicts of interest to disclose. Supported by Georgia Regents University, Augusta, GA. Correspondence: Thomas A. Dillard, MD, Georgia Regents University, 1120 15th Street, BBR-5513, Augusta, GA 30912 (E-mail: [email protected]).
The American Journal of the Medical Sciences
imaging within the study. In most cases, the tissue diagnosis of cancer was made 3 to 5 years after NLST study screening phase ended. Actions such as diagnostic imaging or biopsy followed by usual treatment were taken as indicated by the imaging findings during the study. The 2011 report cited 356 deaths in the LDCT group and 443 deaths in the XRAY group attributable to lung cancer as determined by the end point verification team, a difference of 87 fewer deaths. The difference was 76 in the death certificate analysis. Other causes of death, such as other cancer, cardiovascular disease, respiratory illness and others, accounted for the majority of deaths. The total number of deaths by all causes was reported as 3,856. The NLST included the histologic types of cancers diagnosed during the course of study after randomization to either LDCT or XRAY screening groups. Obviously, tissue type could not be a condition for stratification at randomization before screening; however, histologic type could have a major bearing on outcomes such as mortality within the studied time frame. Depending on stage of disease, patients with squamous cell carcinoma or adenocarcinoma, for example, would typically survive longer than patients with small cell carcinoma and certain other less common histologic types of non-small cell (NSC) lung cancer cited below. This study aimed to compare the proportion of tumor histologic types in the NLST between the LDCT and XRAY screened populations. Given the greater sensitivity of LDCT scanning compared with XRAY, the hypothesis was raised that all histologic types of lung cancer would be detected more often in the LDCT group. The null hypothesis held that the proportion of each tissue type should be equal in the 2 screened groups.
METHODS Data were extracted regarding the tissue types of lung cancer diagnoses from the data published in the NLST report, in particular tables 5, 6 and 7 in that report. Data were excluded from consideration for 8 carcinoid cases in this report, 6 in the LDCT group and 2 in the XRAY group, due to the usual slow growth or benign nature of those tumors. Institutional review board approval was not required for this report. The NLST reported data for tissue types of lung cancers including bronchoalveolar cell carcinomas, adenocarcinomas, squamous cell carcinomas, large cell carcinomas, small cell carcinomas and other non-small cell carcinomas (ONSC). Clinicians have long grouped adenocarcinomas, squamous cell, bronchoalveolar cell and large cell carcinomas together as NSC carcinomas. The ONSC category in the NLST did not include the customary NSC types as noted above. Although the term bronchoalveolar cell carcinoma has been replaced in current nomenclature by the term adenocarcinoma in situ, the NLST considered each of these NSC tissue types separately as has this current report. In the NLST, ONSC carcinomas included the tissue types adenosquamous carcinomas (N 5 28), sarcomatoid carcinomas (N 5 6), unclassified carcinomas (N 5 55), anaplastic
Volume 350, Number 3, September 2015
219
Dillard et al
carcinomas (N 5 1), carcinosarcoma (N 5 1) and those coded only as non-small cell carcinoma (N 5 198). The ONSC tumor breakdown was not reported for each group separately. Many of these tumors have a known poor prognosis.4–10 Data analysis used 2 by 2 contingency tables with x2 statistics and Yates’ correction for continuity (Table 1). Fisher’s exact tests were used where appropriate. The authors report P values less than 0.05 as statistically significant.
RESULTS A total of 1,993 cases of tissue diagnosis of lung cancer were reported in the NLST, 1,054 for the LDCT group and 939 for the XRAY group. Table 1 shows the number of cases of each histologic type in the LDCT and XRAY screened groups of the NLST. The number of small cell carcinomas and ONSC tissue types in the XRAY group was higher than the LDCT screened group (Table 1, Figure 1), and the proportion was statistically significantly greater (P , 0.05). The LDCT group had a significantly larger proportion of bronchoalveolar cell carcinoma (P , 0.05). This is consistent with the greater sensitivity of LDCT for this tissue type. The proportions of adenocarcinomas, squamous cell and large cell carcinomas did not differ statistically significantly between groups (Table 1). Next, the distribution of cancer stages for the small cell and ONSC tumor types was examined. The proportion of stage IV tumors for these 2 tissue types combined was significantly greater in the XRAY (43 more cases) than the LDCT group (P , 0.05). Considered separately, the amount was significant for ONSC (P , 0.05; Figure 2).
DISCUSSION This focused review found a higher proportion of 2 adverse histologic types in the XRAY group of the NLST, small cell and ONSC carcinomas. In addition, a higher proportion of stage IV tumors with these tissue types was observed in the XRAY group. The disproportion of stage IV small cell and ONSC tumors in the XRAY group tumors also supports the contention of adverse biologic nature of these tissue types. These results run counter to the expectation of greater sensitivity with LDCT scanning. These also run counter to the null hypothesis that the default distribution of tissue types
FIGURE 1. Histologic types of lung cancer by screened group in the NLST. The proportion of BAC cases was significantly greater in the LDCT group. The proportion of ONSC and small cell cases in the XRAY group was statistically significantly higher. The other tissue types did not reach statistical significance. BAC, bronchoalveolar cell carcinoma; LDCT, low-dose computed tomography; NLST, National Lung Screening Trial; ONSC, other non-small cell.
would be equal in both groups. As stated above, tissue type could not be a condition for randomization before screening; however, the uneven results of tissue types have some implications on the apparent results of screening. One implication is the potential for length time bias in the NLST.11 Length time bias occurs when the disease severity is worse for one group compared with another. Aggressive tumors have a shorter natural history, leading to earlier death, than less aggressive tumors. Higher mortality in the XRAY group due to greater severity of disease would falsley increase the apparent mortality benefit of LDCT scanning. This effect confounds the interpretation of the lower mortality in the LDCT group as solely due to superior detection of lung cancers and earlier intervention. The foregoing discussion suggests that there may be less mortality benefit from LDCT screening than first reported; however, this current review considers only a single aspect of
TABLE 1. Histology of lung cancer diagnoses in the National Lung Screening Trial by screened group LDCT XRAY All Carcinomas BAC Adeno Squamous Large ONSC Small cell Unknown All
N
%
N
110 380 243 41 131 137 12 1,054
10.4 36.1 23.1 3.9 12.4 13 1.1 100
35 328 206 43 158 159 10 939
%
N
3.7 145 34.9 708 21.9 449 4.6 84 16.8 289 16.9 296 1.1 22 100 1,993
%
P
7.3 35.5 22.5 4.2 14.5 14.9 1.1 100
0.0001* 0.6064; NS 0.5552; NS 0.5031; NS 0.0061* 0.0139* 1.0000; NS —
Data for histology was taken from table 6 of the NLST. BAC, bronchoalveolar cell carcinoma; ONSC, other non-small cell carcinoma. * indicates P , 0.05
220
FIGURE 2. Stage at presentation of ONSC tumors in the NLST by screened group in the NLST. The XRAY group had significantly more stage IV ONSC tumors. The LDCT group had significantly more stage 1A tumors. Other stages did not reach statistical significance. LDCT, low-dose computed tomography; NLST, National Lung Screening Trial; ONSC, other non-small cell. Volume 350, Number 3, September 2015
Uneven Distribution of Cancer Histology
the NLST data and makes no attempt to refute a net benefit of screening. One purpose of this report is to elicit further discussion on the uneven distribution. In summary, this review performed a focused analysis of the distribution of cancer tissue types in the NLST. The main significance of this report is to call attention to the uneven distribution of tissue types, which have worse prognosis. ACKNOWLEDGMENTS The authors thank the National Cancer Institute for access to NCI’s data collected by the National Lung Screening Trial. The statements contained herein are solely those of the authors and do not represent or imply concurrence or endorsement by NCI. REFERENCES 1. Available at: http://www.uspreventiveservicestaskforce.org/uspstf/ uspslung.htm. Accessed January 8, 2014. 2. National Lung Screening Trial Research Team. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395–409. 3. Bach PB, Mirkin JN, Oliver TK, et al. Benefits and harms of CT screening for lung cancer: a systemic review. JAMA 2012;307: 2418–29.
Copyright © 2015 by the Southern Society for Clinical Investigation.
4. Available at: http://seer.cancer.gov/csr/1975_2010/browse_csr.php? sectionSEL515&pageSEL5sect_15_table.13.html. Accessed January 10, 2014. 5. Lee Y, Chung JH, Kim SE, et al. Adenosquamous carcinoma of the lung: CT, FDG PET, and clinicopathologic findings. Clin Nucl Med 2014;39:107–12. 6. Sökücü SN, Kocatürk C, Urer N, et al. Evaluation of six patients with pulmonary carcinosarcoma with a literature review. Scientific WorldJournal 2012;2012:167317:1–5. 7. Filosso PL, Ruffini E, Asioli S, et al. Adenosquamous lung carcinomas: a histologic subtype with poor prognosis. Lung Cancer 2011;74: 25–9. 8. Yendamuri S, Caty L, Pine M, et al. Outcomes of sarcomatoid carcinoma of the lung: a surveillance, epidemiology, and end results database analysis. Surgery 2012;152:397–402. 9. Venissac N, Pop D, Lassalle S, et al. Sarcomatoid lung cancer (spindle/giant cells): an aggressive disease? J Thorac Cardiovasc Surg 2007;134:619–23. 10. Martin LW, Correa AM, Ordonez NG, et al. Sarcomatoid carcinoma of the lung: a predictor of poor prognosis. Ann Thorac Surg 2007;84: 973–80. 11. Available at http://www.acponline.org/clinical_information/journals_ publications/ecp/marapr99/primer.htm Accessed January 15, 2014.
221