Limitation of Annual Screening Chest Radiography for the Diagnosis of Lung Cancer

Transcription

1 2341 Limitation of Annual Screening Chest Radiography for the Diagnosis of Lung Cancer A Retrospective Study Hiroshi Soda, M.D.," Hiroshi Tomita, M.D.,t Shigeru Kohno, M.D.,* and Mikio Oka, M.D.* Background. The incidence and mortality rates of lung cancer have been increasing in many countries. However, the true effectiveness of screening for lung cancer is still controversial. This study aimed to examine the growth pattern of lung cancer and to evaluate the efficacy of screening. Methods. The authors linked the records of annual radiologic screening to cancer registry data and conducted a retrospective follow-up study of radiographs in all patients with lung cancer arising in a population undergoing screening. Results. Among a total of 305,934 participants, screening detected 206 lung cancers, 103 of which were Stage I disease. Seventy-one of the 131 adenocarcinomas were Stage I, and 58% of them showed evidence of cancer for 2 years on a retrospective review of radiographs. Presentation as small faint lesions overlapping the normal chest structures delayed the early detection of adenocarcinoma. The overall sensitivity of screening was 70%, 52% for squamous cell carcinoma and 50% for small cell carcinoma. Rapidly growing Stage 11-IV tumors without retrospective evidence of cancer on previous radiographs accounted for most of the cancers detected during the intervals between screening. Conclusions. Both the low detectability of Stage I adenocarcinoma and the late recognition of rapidly growing small cell and squamous cell carcinomas reduced the effectiveness of screening. More effective imaging methods From the *Second Department of Internal Medicine, Nagasaki University School of Medicine, Nagasaki, Japan, and tnagasaki General Health Center, Nagasaki, Japan. The authors thank Dr. Kohei Hara for helpful advice and Dr. Midori Soda, Radiation Effect Research Foundation, and Dr. Tetsuro Kanda, National Ureshino Hospital, for data collection. Address for reprints: Hiroshi Soda, M.D., Second Department of Internal Medicine, Nagasaki University School of Medicine, 7-1 Sakamoto-machi, Nagasaki 852, Japan. Accepted for publication May 17, and an antismoking campaign are required to reduce lung cancer mortality. Cancer 1993; 72: Key words: lung neoplasms, radiography, tumor growth, screening. The incidence and mortality rates of lung cancer have been increasing in many countries around the world. Although chest radiography and sputum cytology were expected to detect curable early lung cancer, several randomized, controlled trials of such screening have shown that they failed to reduce mortality rates.'-4 Routine screening for lung cancer is, therefore, not recommended at present.' However, the true effectiveness of screening is still contr~versial,~*' and few studies have clarified why screening did not change the natural history of lung We linked the records of radiologic screening to cancer registry data and conducted a retrospective follow-up study of roentgenograms in patients with lung cancer. This study aimed to both examine the growth pattern of lung cancer and to evaluate the usefulness of radiologic screening. Materials and Methods The study consis'ted of two parts. In Part 1, Stage I lung cancers detected by radiologic screening were investigated. In Part 2, lung cancers detected during the intervals between screening tests were analyzed. The central type of squamous cell carcinoma, which has been usually detected by sputum cytology only, was excluded from these analyses. Part 1 In Nagasaki prefecture of Japan, the Nagasaki General Health Center has conducted annual mass screening

2 2342 CANCER October 25, 2993, Volume 72, No. 8 radiologic surveys for lung cancer since 1985, using the existing screening system for tuberculosis. The target population was all residents 40 years of age or older and of both sexes, regardless of smoking habits in Nagasaki prefecture. Although the target population was a total of 881,273 persons, 305,934 persons who consisted of 112,415 males and 193,519 females actually underwent screening for 3 years ( ). The participants corresponded to approximately 35% of the target population. Seventy-eight percent of the participants underwent serial screenings, and the remainder of them underwent only a single radiograph. The studies of reviewing the previous radiographs included 78% of the participants, and the other studies included the entire group of participants. A 100-mm indirect radiograph in posteroanterior view was taken with a mirror camera. The energy of the X-ray beam was 140 kvp, and the focused type grids (ratio, 1O:l; density, 34 lines/cm) were used to eliminate scatter radiation. A rare earth-intensifying fluorescent screen was employed to decrease X-ray exposure and improve the image. The fluorescent image was intensified through a mirror lens and photographed on a 100-mm-sized ortho-type film. This screening method has been the standard mass screening method in Japan because of its low cost and reduced manpower needs. Recently, two comparative studies of 100-mm indirect radiography and a conventional one have been reported in Japan.11r12 These studies are conducted by receiver operating characteristics analysis for detection of a simulated lung nodule and small-sized lung cancer. There are no significant differences between indirect radiography and a conventional one in the area under receiver operating characteristics curve. Their results show that 100-mm indirect radiography is almost equivalent to full-sized direct radiography for detection of lung nodules. Two pulmonary physicians independently interpreted the films with a magnifying glass and compared them with previous ones when any abnormalities were found. The patients with lesions suggestive of cancer were referred to hospitals for diagnostic and therapeutic procedures. Conventional chest radiography and tomography, chest-computed tomography if necessary, were undergone to exclude other pulmonary diseases or artifacts. When the lesion was still suspicious for cancer, bronchoscopic procedures, transcutaneous needle aspiration, or exploratory thoracotomy was performed for accurate diagnosis. Histologically confirmed lung cancer was classified according to the World Health Organization histologic clas~ification, ~ and the International Union Against Cancer staging system.14 Staging procedures included computed tomography, magnetic resonance imaging, and radionuclide scan, for example. In patients with Stage I disease, the radiographs of the past 3 years were reviewed and compared to the findings at the time of detection. Part 2 In Nagasaki, the Radiation Effect Research Foundation has registered patients with lung cancer since The reliability of cancer registry was confirmed by the proportion of cases registered by death certificates only and the ratio of incidence to cancer deaths, 14.9% and 1.7, re~pectively. ~ The lung cancer registry was linked to the records of a total of 205,401 participants who were screened during 1987 and The medical information in the hospitals about the patients was reviewed, such as diagnosis date, cancer location, histologic types, and stage. Histologic types and stage were confirmed by the same approach as described in Part 1. The individuals with lung cancer were divided into screening-detected and interval cases. The former group were the individuals detected by routine annual screening, while the latter group were the individuals diagnosed during the intervals between screening because of symptoms or incidental radiographs. An estimate of the real sensitivity of mass screening is quite difficult. In this study, the program sensitivity of Morrison16 was used: false-negatives were interval cases and false-positives were cases of other diseases or artifacts with a positive screening result. The sensitivity and the specificity were calculated on the basis of these definitions. For example, the sensitivity was defined as follows: sensitivity = (number of screening-detected cases)/(total number of lung cancers arising in the study population). The radiographs obtained 1 year before diagnosis were reviewed in both screening-detected and interval cases. Rapidly growing tumors were defined as Stage 11-IV disease without evidence of cancer on the previous radiographs when reviewed retrospectively. We calculated the proportion of rapidly growing tumors among all lung cancers as follows: proportion of rapidly growing cancers = (the proportion among screeningdetected cases X sensitivity) + (the proportion among interval cases X [lo0- sensitivity]). Statistical Analysis The results for both parts of the study were evaluated by the chi-squared test and Wilcoxon rank-sum test. A P < 0.05 was considered to indicate statistical significance.

3 Screening for Lung Cancer/Soda et al. Results Part 1 Table 1 lists the proportion of Stage I cancers. Fifty percent of the tumors detected were Stage I. Stage I small cell carcinoma was less common than Stage I adenocarcinoma (P < 0.01) and squamous cell carcinoma (P < 0.05). All Stage I small cell carcinomas presented as a peripheral solitary nodule on radiography. Table 2 lists the size of the Stage I adenocarcinomas and squamous cell carcinomas. Stage I adenocarcinoma was smaller in diameter than Stage I squamous cell carcinoma (P < 0.001), and Stage I cases showed a significant decrease among adenocarcinomas larger than 2.0 cm in diameter (P < ). Table 3 lists the retrospective findings obtained from the previous radiographs of patients with Stage I adenocarcinomas and squamous cell carcinomas. Most patients with Stage I adenocarcinoma had retrospective evidence of cancer for several years before they were actually detected. The faintness of lesions superimposed on chest structures, such as the ribs, caused failure to identify them when they first appeared. In contrast, patients with Stage I squamous cell carcinoma showed no evidence of cancer on radiographs taken 2 or 3 years before detection. Part 2 The number positive to the screening test was 1059 cases, containing 140 screen-detected cases (true-positives) and 919 false-positives, The number negative to the screening test was 204,342 cases, containing 59 interval cases (false-negatives) and 204,283 true-negatives. The sensitivity of screening was 70% (140/ ), and the specificity was 99% (204, 283/204, ). The sensitivity of screening according to histologic diagnosis is listed in Table 4. The sensitivity of Table 1. Proportion of Stage I Tumors Among All Lung Cancers Detected by Screening in Relation to Histologic Diagnosis* Histologic diagnosis Total No. Stage I (YO) Adenocarcinoma (54)t Squamous cell (56)$ Small cell 17 3 (18) Large cell 12 4 (33) Other 3 2 (67) Overall (50) * P values were determined by the chi-square test t P < 0.01 compared with small cell carcinoma. P < 0.05 compared with small cell carcinoma. Table 2. Diameter of Stage I Adenocarcinoma and Squamous Cell Carcinoma Adenocarcinoma 2343 Squamous cell carcinoma Tumor diameter* (cm) 2.3 * l.ot 3.4 k 1.5 Proportion of Stage I disease versus tumor size (YO) 5 2 cm 32/38 (84)$ 4/6 (67) cm 22/37 (60) 8/9 (89) z 3.1 cm 17/56 (30) 12/28 (43) * Mean * standard deviation. P < compared with Stage I squamous cell carcinoma by Wilcoxon ranksum test. $ P < 0,0001 for the relationship between tumor size and the proportion of Staee I adenocarcinoma bv the chi-sauare test. squamous cell carcinoma and small cell carcinoma was lower than that for adenocarcinoma (P < ; P < 0.001, respectively). Table 5 lists the proportion of Stage I cancers among screening-detected and interval tumors. In contrast to screening-detected lung cancer, the proportion among interval tumors was quite low (P < ). The proportion of rapidly growing tumors is listed in Table 6. We reviewed 106 previous radiographs of 140 screening-detected cases and the radiographs from 59 interval cases. The proportion of rapidly growing tumors was higher among interval cases than among screening-detected cases (P < 0,0001). Small cell and squamous cell carcinomas were more often rapidly growing tumors than adenocarcinomas (P < ; P < 0.001, respectively). Discussion Lung cancer is a highly lethal malignancy, and only resected Stage I patients have a reasonable probability of cure with a 5-year survival rate of 65%.17 The main purpose of screening for lung cancer is, therefore, to detect more Stage I cases, thus reducing the mortality rate. Earlier radiologic screening studies have detected approximately 50-70% of lung cancers arising in the study populations, and generally about half of the detected lung cancers are Stage The proportion of Stage I disease is, thus, not large enough to decrease the mortality of the screened population. Several randomized, controlled trials with radiography and sputum cytology have failed to reduce the mortality.'-* Even in two case-control studies for comparing screening with no screening, the benefit of screening has not been demonstrated ~learly,~,~ although one of the studies showed statistically borderline benefit.7 We considered that the following two factors might be acting to reduce the pro-

4 2344 CANCER October 25, 2993, Volume 72, No. 8 Table 3. Retrospective Evidence of Cancer on Previous Radiographs in Patients With Stage I Adenocarcinoma and Squamous Cell Carcinoma* Time Adenocarcinoma Squamous cell carcinoma before detection No. No. with evidence No. No. with evidence (yr) reviewed of cancer (YO) reviewed of cancer (YO) (85) 21 9 (43)t (58) 18 0 (OH (29) 14 0 (0) * P values were determined by the chi-square test. t P < compared with Stage I adenocarcinoma. $ P < compared with Stage I adenocarcinoma. portion of Stage I disease detected by radiologic screening: (1) a low level of detectability of Stage I adenocarcinoma by roentgenography, which is a serious problem, especially in East Asia with a high frequency of adenocarcinoma",20 and (2) a high proportion of rapidly growing cancers, especially small cell and squamous cell carcinomas. Adenocarcinoma grows more slowly than other types of lung cancer.' The mathematical model of Flehinger suggests that the mean duration from Stage I to advanced disease is at least 4 years and may be considerably longer.' However, the mathematical model also indicates that the detectability of Stage I adenocarcinoma is less than 16% by a single radiologic screening procedure.' Most Stage I adenocarcinomas are retrospectively shown to be present on earlier roentgenograms for several years before they are actually detected, as listed in Table 3.2',22 It is quite difficult to detect adenocarcinoma at its first radiologic presentation. Pa thologically, adenocarcinoma replaces the alveolar cells and invades the surrounding lung parenchyma, while initially preserving the air spaces. As the carcinoma develops, central fibrosis in the tumor causes the bronchi, vessels, and pleura to converge toward the le- Table 4. Sensitivity of Radiologic Screening According to Histologic Diagnosis* Histologic Screening- Interval Sensitivity diagnosis detected cases cases (YO) Adenocarcinoma Squamous cell t Small cell $ Large cell Other Overall * P values were determined by the chi-square test t P < compared with adenocarcinoma. t P < cornoared with adenocarcinoma. sion.23,24 This process results in a fairly small tumor, even at an advanced stage of the disease. Radiologically, early adenocarcinoma appears as a localized faint infiltration rather than as a nodular lesion.' In addition to its radiologic characteristics, namely a small-sized nodule with ill-defined margin and low density, overlapping with normal chest structures tends to delay the early detection of these tumors.25 The smallest cancers that are prospectively detectable by radiography are approximately 1.O cm in diameter,25 and adenocarcinomas larger than 2.0 cm in diameter readily metastasize to lymph nodes or other distant organs.26 As listed in Table 2, 58% of adenocarcinoma larger than 2.0 cm already metastasized at their detection. Therefore, adenocarcinoma has to be detected while it presents as a faint lesion ranging from cm in diameter. More sensitive imaging methods may be required as screening tools, because standard roentgenography cannot easily detect such small faint lesions as previously mentioned. Among new radiologic techniques including computed tom~graphy,~~-~' spiral volumetric computed tomography seems to be more attractive, which is equivalent to conventional computed tomography for detection of a small nodule.30 If this method is used for screening tool, we have to solve many practical problems, including the risk of radiation exposure, cost, and manpower. The new radiologic methods for screening should be effective, efficient, and available. The peripheral type of squamous cell carcinoma grows expansively in contrast to adenocarcinoma, with Table 5. Proportion of Stage I Cancers Among Screening-Detected and Interval Cases Screening-detected Interval cases cases All cases State I (YO) 73 (52) 7 (12)* 80 (40) Stage 11-1V (YO) 67 (48) 52 (88) 119 (60) Overall * P < comdared with screenine-detected cases bv the chi-sauare test.

5 Screening for Lung Cancer/Soda et al Table 6. Proportion of Rapidly Growing Tumors According to Histologic Diagnosis* Screening-detected cases Interval cases All cases Histologic Rapidly Rapidly Rapidly diagnosis Reviewed growing (Yo) Reviewed growing (Yo) growing (Yo) Adenocarcinoma 65 5 (8) 16 7 (43) 13 Squamous cell 25 7 (28) (50) 39t Small cell 6 5 (83) (83) 83$ Large cell 8 1P3) 3 2 (67) 27 Other 2 0 (0) 4 0 (0) 0 Overall (17) (53) 27 * P values were determined by the chi-square test. t P < compared with adenocarcinoma. $ P < compared with adenocarcinoma. P < 0.01 compared with squamous cell carcinoma. 5 P < compared with screening-detected cases. filling alveolar spaces and displacing the surrounding normal structures. This process radiologically shows moderately well-defined margin and homogeneous density, which appear to be detected more easily than adenocarcinoma. However, its sensitivity of radiologic detection is lower than that of adenocarcinoma,18 because this type of squamous cell carcinoma develops moderately rapidly. The sensitivities in this study were 52% in squamous ceil carcinoma and 85% in adenocarcinoma (Table 4). Even though a recent radiograph has showed normal findings, Stage I squamous cell carcinoma larger than 3.0 cm in diameter can appear within 1 year, as listed in Tables 2 and 3. Thus, annual screening might not detect more Stage I squamous cell carcinomas even if the radiographs were interpreted carefully. The present study did not clarify whether screening at a shorter interval could detect more Stage I cases, but population screening at shorter intervals than 1 year cannot be recommended because of manpower limitations and high costs. The peripheral type of squamous cell carcinoma is not a suitable disease to detect by screening because of its moderately rapid growth. Small cell carcinoma grows very rapidly and metastasizes early, so Stage I tumors are uncommon. Small cell carcinoma is usually located proximally, but the Stage I disease presented as single peripheral nodules in this study. It is debatable whether peripheral small cell carcinoma differs from central type small cell carci- noma and atypical ~arcinoid.~l,~* However, the peripheral type is quite rare, and small cell carcinoma is not a suitable disease to detect by screening. Cigarette smoking is a major cause of lung cancer, especially of small cell and squamous cell carcinoma^.^^ The risk of developing lung cancer is determined by the number of cigarettes smoked, the age when smoking begins, and the degree of inhalation. However, the risk decreases rapidly over 10 years after cessation of smoking.33,34 Small cell and squamous cell carcinomas are not suitable for screening because of their rapid growth, so the cessation of smoking appears to be the best method of reducing the incidence of these cancers. In conclusion, none of the histologic types of lung cancer are effectively detected by the present method of radiologic screening. More sensitive imaging methods for adenocarcinoma and an antismoking campaign for small cell and squamous cell carcinomas are required to reduce lung cancer mortality. References 1. Fontana RS, Sanderson DR, Woolner LB, Taylor WF, Miller WE, Muhm JR, et al. Screening for lung cancer: a critique of the Mayo lung project. Cancer 1991; 67: Melamed MR, Flehinger BJ. Detection of lung cancer: highlights of the Memorial Sloan-Kettering Study in New York City. Schweiz Med Wochenschr 1987; 117: Kubik A, Parkin DM, Khlat M, Erban 1, Polak J, Adamec M. Lack of benefit from semi-annual screening for cancer of the lung: follow-up report of a randomized controlled trial on a population of high-risk males in Czechoslovakia. Int J Cancer 1990; 45: Tockman MS. Survival and mortality from lung cancer in a screened population: the Johns Hopkins study. Chest 1986; 89(S~ppl): Eddy DM. Screening for lung cancer. Ann Intern Med 1989; 111: Berndt R, Nischan P, Ebeling K. Screening for lung cancer in the middle-aged. Int J Cancer 1990; 45: Sobue T, Suzuki T, Naruke T, the Japanese-lung-cancer screening research group. A case-control study for evaluating lungcancer screening in Japan. Int J Cancer 1992; 50: Flehinger BJ, Kimmel M, Melamed MR. Natural history of adenocarcinoma-large cell carcinoma of the lung: conclusions from screening programs in New York and Baltimore. ] Natl Cancer Inst 1988; 80: Weiss W, Boucot KR, Seidman H. The Philadelphia pulmonary neoplasm research project. Clin Chest Med 1982; 3: Nakada T, Sato H, Saito Y, Fujimura S. Detection of early lung cancer: results of radiologic and cytologic screening in the Miyagi program. Tohoku Exp Med 1987; 152: Itoh H, Takashima T, Kamimura R, Nagatoh H, Nagata K. (Unpublished observations) Proceedings of the Twenty-Seventh An-

6 2346 CANCER October 15, 1993, Volume 72, No nual Meeting of the Japan Lung Cancer Society; 1986 Oct 30-31; Tokyo, Japan. Kamimura R, Kobayashi T, Arakawa F, Ueda T, Takanaka T, Nishijima H, et al. (Unpublished observations) Proceedings of the Thirty-second Annual Meeting of the Japan Lung Cancer Society; 1991 Oct 30-31; Ohtsu, Japan. World Health Organization. The World Health Organization histological typing of lung tumors. ed. 2. Am J Clin Pathol 1982; 77: Mountain CF. A new international staging system for lung cancer. Chest 1986; 89(Suppl): Nagasaki prefectural cancer registry office: Nagasaki prefectural cancer registry operation report (in Japanese). Nagasaki: Radiation Effect Research Foundation, Morrison AS. Screening in chronic disease. New York: Oxford University Press, 1985: Naruke T, Goya T, Tsuchiya R, Suemasu K. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988; 96: Sobue T, Suzuki T, Matsuda M, Horai T, Kajita A, Kuriyama K, et al. Sensitivity and specificity of lung cancer screening in Osaka, Japan. Jpn J Cancer Res 1991; 82: Watanabe S, Tsugane S, Arimoto H, Shimozato Y, Suemasu K, Arai H, et al. Trend of lung cancers in the National Cancer Center of Japan and comparison with that of Japanese pathological autopsy records. Jpn J Cancer Res 1987; 78: Lei YX, Du YX, Chen YZ. Analysis of the relationship between lung cancer cell type and some associated factors. Lung Cancer 1991; 7(Suppl): 2. Heelan RT, Flehinger BJ, Melamed MR, Zaman MB, Perchick WB, Caravelli JF, et al. Non-small-cell lung cancer: results of the New York screening program. Radiology 1984; 151: Muhm JR, Miller WE, Fontana RS, Sanderson DR, Uhlenhopp MA. Lung cancer detected during a screening program using four-month chest radiographs. Radiology 1983; 148: Barsky SH, Huang SJ, Bhuta S. The extracellular matrix of pulmonary scar carcinomas is suggestive of a desmoplastic origin. Am JPathol 1986; 124: Suzuki A. Growth characteristics of peripheral type adenocarcinoma of the lung in terms of roentgenologic findings. In: Shimozato Y, Melamed MR, Nettesheim P, editors. Morphogenesis of lung cancer. vol. 1. Florida: CRC Press, 1982: Woodring JH. Pitfalls in the radiologic diagnosis of lung cancer. AIR 1990; Sagawa M, Saito Y, Takahashi S, Usuda K, Kamma K, Sato M, et al. Clinical and prognostic assessment of patients with resected small peripheral lung cancer lesions. Cancer 1990; 66: Kuriyama K, Tateishi R, Doi 0, Kodama K, Tatsuta M, Matsuda M, et al. CT-pathologic correlation in small peripheral lung cancers. AIR 1987; 149: Zwirewich CV, Veda1 S, Miller RR, Muller NL. Solitary pulmonary nodule: high-resolution CT and radiologic-pathologic correlation. Radiology 1991; 179: Vock P, Soucek M, Daepp M, Kalender WA. Lung: spiral volumetric CT with single-breath-hold technique. Radiology 1990; 176~ Costello P, Anderson W, Blume D. Pulmonary nodule: evaluation with spiral volumetric CT. Radiology 1991; 179: Gephardt GN, Grady KJ, Ahmad M, Tubbs RR, Mehta AC, Shepard KV. Peripheral small cell undifferentiated carcinoma of the lung: clinicopathologic features of 17 cases. Cancer 1988; 61: Quoix E, Fraser R, Wolkove N, Finkelstein H, Kreisman H. Small cell lung cancer presenting as a solitary pulmonary nodule. Cancer 1990; 66: Lubin JH, Blot WJ. Assessment of lung cancer risk factors by histologic category. J Natl Cancer Insf 1984; 73: Sobue T, Suzuki T, Fujimoto I, Matsuda M, Doi 0, Mori T, et al. Lung cancer risk among exsmokers. Jpn J Cancer Res 1991; 82:273-9.