J Cancer Res Clin Oncol (2003) 129: 114–122 DOI 10.1007/s00432-002-0408-4

O R I GI N A L P A P E R

Branislav Jeremic Æ Biljana Milicic Aleksandar Dagovic Æ Jasna Aleksandrovic Nebojsa Nikolic

Pretreatment clinical prognostic factors in patients with stage IV non-small cell lung cancer (NSCLC) treated with chemotherapy Received: 27 September 2002 / Accepted: 21 November 2002 / Published online: 7 March 2003
Springer-Verlag 2003

Abstract Purpose: We investigated the influence of potential pre-treatment clinical prognostic factors in stage IV non-small cell lung cancer (NSCLC). Methods and patients: A total of 285 patients were enrolled in two consecutive prospective randomised studies which compared (study 1) carboplatin and prolonged oral etoposide (group 1; n=58) with the same etoposide alone (group 2; n=59), and (study 2) carboplatin and prolonged oral etoposide (group 1; n=84) with the same carboplatin and high-dose intravenous etoposide (group 2; n=84). Results: The median survival time for all 285 patients was 7 months, while 1- and 2-year survival rates were 29% and 8%, respectively. Age did not impact on outcome (P=0.21), while female patients did significantly better than male patients (P<0.0001). Patients with KPS 80–100 did significantly better than those with KPS 50–70 (P<0.0001), as did patients with less pronounced weight loss (P<0.0001) and those with only one metastatic site when compared to those having at least two metastatic sites (P<0.0001). When evaluated regarding the metastatic site, only subcutaneous metastatic site did not influence survival. This was confirmed within univariate analyses, but when multivariate analyses were done gender, KPS, weight loss, number of metastatic sites, presence of liver metastases and presence of brain metastases independently influenced survival, while age and other metastatic locations did not. Conclusion: In this analysis, gender, KPS, weight loss, number of metastatic sites, presence of liver metastases and presence of brain metastases independently influB. Jeremic (&) Department of Radiotherapy, Klinikum rechts der Isar, Technical University Munich, Ismaninger Strasse 22, 81675 Munich, Germany E-mail: [email protected] Tel.: +49-89-41404512 Fax: +49-89-41404882 B. Milicic Æ A. Dagovic Æ J. Aleksandrovic N. Nikolic Æ B. Jeremic Department of Oncology, University Hospital, Kragujevac, Yugoslavia

enced survival in patients with stage IV NSCLC treated with CHT. Keywords Non-small cell lung cancer Æ Chemotherapy Æ Prognostic factors

Introduction Lung cancer is the commonest cause of cancer death in the western world. At least two-thirds of cases are nonsmall cell lung cancer (NSCLC) and approximately half of these patients present with metastatic (stage IV) disease [1]. The ‘‘optimal’’ treatment approach to this disease remains controversial. Disappointing results obtained with chemotherapy (CHT) in stage IV NSCLC with accompanying toxicity have led many to carry out investigations which tested it against best supportive care (BSC). The results of three recent meta-analyses showed a reduction in mortality during the first 6 months for patients treated with combination CHT with the overall improvement in median survival time (MST) of only about 3 months [2]. CHT was associated with a 24% reduction in the probability of death when compared to BSC, but this effect decreased significantly after the first 6 months from the beginning of therapy. The mean potential gain in survival/overall net benefit for CHT as compared with BSC was approximately 6 weeks [3, 4]. Another unsettled issue in the ‘‘optimal’’ treatment of stage IV NSCLC is whether combination CHT is better than single-agent CHT. Two recent meta-analyses of single-agent versus combination CHT in stages III and IV NSCLC showed that combination CHT increased objective response rate as well as toxicity [5, 6]. However, survival was only modestly prolonged with combination CHT, but not significantly so when more active single agents were used. While all these studies used data available from studies which used ‘‘second’’ generation, mostly cisplatin

115

(CDDP)-based, CHT, the last decade witnessed a widespread introduction of ‘‘third’’ generation drugs such as vinorelbine, taxanes, camptothecins, and gemcitabine. They all showed promising results when used alone in advanced NSCLC, but better results were obtained when these ‘‘new’’ drugs were combined with either CDDP or carboplatin (CBDCA) in numerous recent prospective randomised trials [7, 8, 9, 10, 11], although not all studies have confirmed this [12, 13]. Additionally, a similar outcome for various ‘‘doublet’’ CHT regimens was accompanied with a different toxicity profile [14, 15], as recent results from the Eastern Cooperative Oncology Group (ECOG) study E1594 [16] have shown. Whichever CHT is used, identification of potential prognostic factors is of paramount importance. A number of pre-treatment clinical and biological factors have been studied over the years. While evaluation of biological pre-treatment prognostic factors is under extensive investigation, clinical pre-treatment factors, both patient- and tumour-related, have been continuously investigated over decades, sometimes with conflicting results. In an attempt to contribute to the ongoing controversy, we therefore undertook an analysis of potential clinical pre-treatment prognostic factors in patients with metastatic (stage IV) NSCLC treated with CHT during two consecutive prospective randomised trials [17, 18].

Material and methods During a 7-year period (1988–1994), two consecutive prospective randomised studies were performed at our institution [17, 18]. To be eligible for these studies, adult patients had to have histologically or cytologically proven stage IV NSCLC according to the International System, Karnofsky performance status (KPS) =50, measurable or evaluable disease, normal haematological (WBC =4000/ll, platelets =100,000/ll, hemoglobin =10 g/dl), hepatic (serum bilirubin <1.5 mg/dl), and renal (serum creatinine <1.5 mg/dl) function, a life expectancy of at least 3 months, and no prior therapy. Patients with recent cardiac disease, such as myocardial infarction or uncontrolled congestive heart failure, were excluded as well as those with a history of any prior or concurrent cancer (except that of the skin), unless the patient had shown no evidence of disease for more than 5 years. Staging procedures included medical history, physical examination, complete blood count with differentials, biochemistry and electrolyte panel, chest X-ray (PA and laterals), and bone marrow examination. Liver and bone radionuclide scan, liver ultrasound, and bone radiographs were performed if needed, while computed tomography (CT) scan of the thorax and upper abdomen were strongly recommended and performed as mandatory during the latter part of study 1 while they were mandatory in all patients in study 2. CT scan of the brain was strongly recommended and performed whenever needed. Treatment was administered in the study 1: group I – CBDCA, 400 mg/m2, day 1 and oral etoposide, 50 mg/m2, days 1–21, and group II – the same etoposide as in group I alone [17]. This study ran from January 1988 through December 1990. In study 2 [18], treatment included: group I – CBDCA, 400 mg/m2, day 1 and etoposide, 50 mg/m2, days 1–21, and group II – the same CBDCA as in group I and etoposide, 200 mg/m2, days 1–3. Etoposide was administered either orally in the form of soft gelatin capsule (group I) or as short i.v. infusion (group II). This study ran from January

1991 through December 1994. In both studies, CBDCA was administered as 30’ i.v. infusion, while oral etoposide was used in 50 mg capsules, with calculated daily dose as previously described by Hainsworth et al. [19]. In both studies (all four treatment groups), treatment was repeated every 28 days for up to six cycles or until tumour progression was noted. Tumour response was assessed after two cycles of treatment, and after each cycle to the end of treatment thereafter, and at 2-month intervals thereafter. Differences between pairs of groups in patient characteristics were evaluated by the chi-square test or Student’s t-test. Survival was calculated by the Kaplan-Meier method and differences between pairs of groups in survival curves were analysed by the logrank test. Univariate and multivariate Cox proportional hazards model were used with both full and best multivariate model using the SPSS (Version 8.0) statistical package. Cut-off dates for this analysis were the dates of closing both studies, before they were analysed and subsequently published. For the purpose of this analysis, the two databases were combined into one.

Results Clinical pretreatment variables included in this study were: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

Gender Age Karnofsky performance status Pretreatment weight loss Number of metastatic sites Presence of liver metastasis Presence of brain metastasis Presence of bone metastasis Presence of lung metastasis Presence of subcutaneous metastasis Presence of bone marrow metastasis Presence of other metastasis, being metastases to organs less frequently observed (e.g., adrenals, skin, etc.).

These variables have all been documented in the two studies that served as a basis for this analysis. The results of the two studies have been published in detail [17, 18]. Briefly, in study 1, a two-drug regimen was superior to single-agent oral etoposide (MST, 9 vs 5 months; 1- and 2-year survival rates, 38% and 12% vs 24% and 5%, respectively; P=0.015), while in study 2, there was no difference in survival between oral and intravenous etoposide when both were given with carboplatin (MST, 8 months vs 7 months; 1- and 2-year survival, 35% and 9.5%, vs 31% and 7.1%, respectively; P=0.40). A total of 285 patients were randomised in these two studies (study 1, n=117; study 2, n=168). No patient was lost to follow-up. Patients characteristics are given in Table 1. All clinical pre-treatment variables investigated in this study were balanced between studies/treatment groups. Although a large number of factors were evaluated, there were no missing values for any of these factors. Median survival time (MST) for all 285 patients was 7 months, while 1- and 2-year survival rates were 29% and 8%, respectively (Fig. 1). A total of 24 patients survived for at least 24 months: study 1, group I, n=7;

116 Table 1 Patient characteristics. (KPS Karnofsky performance status score, W. loss weight loss)

Variable

Gender Age (years) KPS W. loss Number of metastatic sites Liver metastases Brain metastases Bone metastases Lung metastases Subcutaneous metastases Bone marrow metastases Other metastases

Study 1

F M <60 ‡60 50–70 80–100 £ 5% >5% 1 ‡2 Yes No Yes No Yes No Yes No Yes No Yes No Yes No

Fig. 1 Overall survival (all patients)

study 1, group 2, n=3; study 2, group 1, n=8; study 2, group 2, n=6). Kaplan-Meier survival analysis showed that although patients ‡60 years old did better than those <60 years old, the influence of age did not reach significance (P=0.21). Female patients did significantly better than male patients (P<0.0001), patients with KPS 80–100 did significantly better than those with KPS 50–70 (P<0.0001), as did patients with only one metastatic site when compared to those having at least two metastatic site (P<0.0001). In addition, an investigation of the

Study 2

Total

Group 1 (n=58)

Group 2 (n=59)

Group 1 (n=84)

Group 2 (n=84)

21 37 40 18 20 38 29 29 40 18 11 47 3 55 22 36 30 28 7 51 2 56 7 51

23 36 37 22 20 39 33 26 45 14 12 47 4 55 16 43 25 34 8 51 3 56 7 52

30 54 50 34 27 57 35 49 57 27 20 64 6 78 26 58 40 44 12 72 5 79 10 74

32 52 46 38 27 57 36 48 59 25 18 66 6 78 26 58 40 44 11 73 7 77 8 76

106 179 173 112 94 191 133 152 201 84 61 224 19 266 90 195 135 150 38 247 17 268 32 253

impact of a particular metastatic site was carried out next. It revealed that the presence of metastasis in the liver (P<0.0001), brain (P<0.0001), bone (P<0.0001), lung (P<0.0001), bone marrow (P<0.0001), and various metastatic sites grouped as ‘‘other’’ (due to the small number of patients in each of these particular categories, such as adrenals, skin, etc.) (P<0.0001) all significantly influence survival. However, the presence of a subcutaneous metastatic site did not influence survival (P=0.55) (Table 2). These results were confirmed when univariate Cox analysis was done. However, when Cox multivariate analysis was done, the best model showed that gender, KPS, weight loss, and number of metastatic sites independently influenced survival, as well as liver and brain metastatic tumour sites, whereas all other variables did not independently influence survival (Table 2). KPS and weight loss were the strongest prognosticators of treatment outcome, followed by, in descending order, the presence of liver metastasis, number of metastatic sites, the presence of brain metastasis, and gender (see Figs. 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13). The possible effect of treatment was then taken into account because in study 1 a significant difference was found between the treatment arms, although this difference in study 2 was not significant. Since the same oral etoposide and standard-dose carboplatin was used in both studies (in study 1 it was arm I and in study 2 it was arm 1), these two arms were combined, making a total of three groups for investigation. The KaplanMeier analysis revealed significant survival advantage for carboplatin and oral etoposide (two groups combined) over oral etoposide alone (P=0.0104). There was no difference between carboplatin and oral etoposide

117 Table 2 Survival analysis according to potentialpretreatment prognostic factors. (KPS Karnofsky performance status score, W. loss weight loss)

Variable

N

MST 1 year 2 years Log-rank Univariate Multivariate Months (%) (%) P P P

Gender

106 179 173 112 94 191 133 152 201 84 61 224 19 266 90 195 135 150 38 247 17 268 32 253

12 6 7 8 3 11 13 5 11 3 4 9 3 8 5 9 10 6 8 7 4 8 3 8

F M Age (years) <60 =60 KPS 50–70 80–100 W. loss =5% >5% No. metastatic sites 1 =2 Liver metastases Yes No Brain metastases Yes No Bone metastases Yes No Lung metastases Yes No Subcutaneous Yes No Bone marrow Yes No Other Yes No

50 17 27 33 3 42 56 6 41 1 0 37 0 31 14 36 41 18 26 30 12 30 9 32

15 4 8 10 0 12 17 1 12 0 0 11 0 9 2 11 14 3 5 9 0 9 3 9

<0.0001

<0.0001

0.0126

0.21

0.23

–

<0.0000

<0.0001

<0.0001

<0.0000

<0.0001

<0.0001

<0.0000

<0.0001

0.0016

<0.0000

<0.0001

0.0013

<0.0000

<0.0001

0.0028

<0.0000

<0.0001

0.07

<0.0000

<0.0001

0.47

0.55

0.57

–

<0.0000

0.0002

0.40

<0.0000

<0.0001

0.53

Fig. 2 Survival according to gender. Females (broken lines) and males (unbroken lines)

Fig. 3 Survival according to age. ‡60 years (broken lines) and <60 years (unbroken lines)

and carboplatin and intravenous etoposide (P=0.22228) and oral etoposide alone and carboplatin and intravenous etoposide (P=0.11). When the Cox univariate model was used, the effect of treatment was not significant (P=0.27) and was not entered into the Cox multivariate model. Therefore, the data in Table 2 remain unchanged. Of the 24 patients who survived ‡2 years, the majority were females (n=16), had an isolated lung metastatic site (n=19), and had no pronounced weight loss (n=22). All of these 24 patients had only one metastatic site and had a KPS of 80–100.

Discussion The current treatment results for stage IV (metastatic) NSCLC are not satisfactory. They call for improvements, which should be related not only to new therapies, but also to identification of patients suitable for a particular treatment approach from which they are most likely to benefit. While various biological characteristics such as genetic (DNA aneuploidy, chromosome abnormality, oncogenes, tumour suppressor genes, etc.), or markers of differentiation, tumour cell proteases, tumour-associated antigens, or other biological

118

Fig. 4 Survival according to KPS. 80–100 (broken lines) and 50–70 (unbroken lines)

Fig. 6 Survival according to the number of metastatic sites. =1 (broken lines) and ‡2 (unbroken lines)

Fig. 5 Survival according to pretreatment weight loss. =5% (broken lines) and >5% (unbroken lines)

Fig. 7 Survival according to the presence of liver metastasis. No presence (broken lines) and presence (unbroken lines)

parameters have been intensively investigated in recent years, there are also numerous pretreatment laboratory (e.g., pretreatment serum LDH, leukocyte count, have been shown to be independent prognosticators in both NSCLC and SCLC) and clinical prognostic factors which must be taken into account. Particularly the latter, since clinical pretreatment prognostic factors are easy to observe/note and they represent the factors which can be collected early in the course of the disease/ diagnosis, well before we decide upon the ‘‘optimal’’ treatment approach in this disease. Although the differences between pretreatment patient and/or tumour characteristics were discussed about 40 years ago [20, 21, 22], it is only in the last two decades that the importance of prognostic factors have been

emphasized. This particularly relates to the design and execution of clinical trials and evaluation of the treatment outcome, since the characteristics of patients can influence the clinical course of the patients [23, 24]. It was observed that the magnitude of differences in outcome for categories of the strongest prognostic factors are larger than those for the type of therapy used [25, 26]. Early studies on prognostic factors in lung cancer suffered from inclusion of all histologies, both ‘‘limited’’ and ‘‘extensive’’ disease, patients having second primary neoplasm or recurrent tumours as well as all treatment approaches [27, 28]. With clinical trials more frequently being practiced in the last two decades, it has become essential to take prognostic factors into consideration to

119

Fig. 8 Survival according to the presence of brain metastasis. No presence (broken lines) and presence (unbroken lines)

Fig. 10 Survival according to presence of lung metastasis. No presence (unbroken lines) and presence (broken lines)

Fig. 9 Survival according to presence of bone metastasis. No presence (broken lines) and presence (unbroken lines)

Fig. 11 Survival according to presence of subcutaneous metastasis. No presence (broken lines) and presence (unbroken lines)

maintain the comparability of separate clinical trials. It is for these reasons, therefore, that the purpose of the analysis of prognostic factors included; 1) selection of important factors which should be used for stratification before randomisation to maintain the comparability of the study groups; 2) characterisation and prediction of treatment outcome; and 3) identification of subgroups that may be given better treatment by different/alternative treatment approaches. With the same aim we performed the current analysis. Although retrospective, it was done within the patient population of exclusive stage IV NSCLC and could be used to generate hypotheses that should result in being validated in a further data set of patients. While one may object to the choice of the drugs we used, particularly

etoposide, now largely replaced by the ‘‘third generation’’ drugs, these studies were designed in mid-1987 (study 1) and mid-1990 (study 2), the time when the use of this drug (and its combination with platinating agents) was widely practiced. In addition, we used only survival as an endpoint because of general problems in investigating progression-free survival, although the latter endpoint may be advantageous in terms of quality of life issues, and which has been considered as a necessary endpoint in studies in recent years. In this analysis we focused on clinical pretreatment factors that are easy to note/score and the monitoring of which can easily be done not only before the treatment, but during the treatment and follow-up period as well. Among these, performance status has been intensively

120

Fig. 12 Survival according to presence of bone marrow metastasis. No presence (broken lines) and presence (unbroken lines)

Fig. 13 Survival according to presence of other metastasis. No presence (broken lines) and presence (unbroken lines)

discussed. Unequivocally, it has been shown that patients having higher performance status score do better than those with lower performance status [27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37], regardless of the scoring system, KPS [30, 31, 32, 33], ECOG [26, 34, 35, 36], SWOG [25, 37]. In contrast to this, Ewans et al. [38] and Sculier et al. [39] did not observe an impact on survival, the latter possibly because of focusing on the best prognostic subgroup (2-year survivors). This prognostic factor is probably the most significant one and is mostly used in the randomisation in clinical studies today. In the current study, KPS was also an independent prognosticator of treatment outcome favouring patients with a KPS of 80–100.

Another factor frequently identified as an independent prognosticator is female gender. There is almost 40 years of evidence that females do better than males in lung cancer studies [20, 21, 22], but this has became even more apparent with an increase in the incidence of lung cancer in females and, consequently, inclusion of more females in clinical studies [25, 26, 29, 31, 33, 34, 36, 37, 40, 41]. While the findings of current study support this observation, there are few studies that did not find and impact of gender on treatment outcome [32, 35, 39, 42]. The factor investigated regarding its effect on treatment outcome is pretreatment weight loss. The results are conflicting: while O’Connell et al. [31], Einhorn et al. [30], Sorensen et al. [32], Shinkai et al. [36], and Sculier et al. [39], did not find that it had an influence on survival, Stanley [28] Finkelstein et al. [26], Rapp et al. [35], and Fukuoka et al. [40] did. A possible explanation for these inconsistencies is at least two-fold. First, this factor has usually been identified as closely correlating with the performance status and second, there were differences among the studies regarding the cut-off values in pounds (10 lb) [28] or kilograms [35] or in percents (5%, 6%, and 10%) [26, 27, 30, 31, 32, 36, 39], inclusion of patients with stage III and stage IV [27, 28, 30, 31, 35, 36, 39] as well as some studies focussing upon long-term survivors as the best subgroup [26, 39]. Additionally, there were differences in the study populations, methodological differences, as well as short follow-up in some studies [31]. At present, this factor should be designated as a possible one. Age was frequently included in the prognostic factor analyses. The vast majority of studies, including the current one, have not shown its impact on survival [25, 26, 29, 31, 32, 35, 36, 39, 40, 42]. Contrary to this, Paesmans et al. [33] and Albain et al. [37] observed its influence. However, while in the study of Albain et al. [37] age =70 years was predictive of improved survival, in the study of Paesmans et al. [33] age of >60 years was associated with relative risk of death of 1.27 (P=0.02). Lanzotti et al. [27] also found age =70 years to be an independent prognosticator of inferior survival in 187 patients with extensive NSCLC. However, Green et al. [43] identified age <50 years as an adverse prognostic factor influencing survival. It is probable that age does not play an independent role in predicting treatment outcome in this patient population, even when palliative CHT is considered [42]. The number of metastatic sites has not frequently been taken into account as a prognosticator of the treatment outcome in this disease. When included within the investigation, it has given contradictory results. In the study of O’Connell et al. [31] the presence of two or more extrathoracic metastatic organ sites predicted shorter survival. Contrary to this, Einhorn et al. [30] did not observe such an effect. However, in a large SWOG study [37] on extensive NSCLC, when disease extent was defined as either a single lesion or multiple lesions, single site or multiple sites, both MST (8.7 vs 6.2 vs 5.1 months) and percent 1-year survival (37% vs 20%

121

vs. 17%), predicted survival in patients with favourable characteristics (PS 0–1) (n=362). A single metastatic site was a prognosticator of improved survival (hazard ratio, 1.6; P=0.02). In addition, Shinkai et al. [36] demonstrated that the number of metastatic sites was the most important prognostic factor of survival in advanced NSCLC and suggested that it should be used as a factor in stratification. Our results are in agreement with those showing its independent influence on survival. Although it seems rational to expect the biggest tumour burden in tumours with multiple metastatic sites and, therefore, somewhat greater task for CHT, new studies and more patients are needed before making definite conclusions about this factor. Information regarding the impact of metastasis to a particular organ on treatment outcome is unclear. Liver metastases were found to be an adverse prognosticator in the studies of Lanzotti et al. [27], Ewans et al. [38], Finkelstein et al. [26], and Sorensen et al. [32] as well as in the current study. This was not so in the studies of O’Connell et al. [31], Miller et al. [25], Sakurai et al. [29], Shinkai et al. [36], Sculier et al. [39], and Hikisch et al. [42], which has made any solid recommendation/conclusions impossible. Brain was found to be adverse metastatic only in the study of Lanzotti et al. [27] and our own, probably due to the small number of patients. While the presence of bone metastases adversely predicted survival in the studies of O’Connell et al. [31] and Finkelstein et al. [26], all other studies [25, 29, 32, 39, 42] as well as our own have not observed this. It is possible that some of the discrepancies in the analyses available in the literature regarding the latter two factors, may at least in part lay in the fact that brain and bone metastases are metastatic sites most frequently treated with palliative radiotherapy which may artificially influence treatment outcome. The same controversy regarding the influence of a particular metastatic site can be extended to other metastatic sites. While metastatic lung site was not usually observed to have an independent influence on survival, including the current one, this was the case in the series of Lanzotti et al. [27]. Subcutaneous metastases, bone marrow metastases, and ‘‘other’’ metastatic sites (grouped as such due to the small number of patients with different metastatic sites such as adrenals, skin, etc.) were mostly found not to predict survival. The results of our study are in agreement with this. Exceptions to this are the studies of Finkelstein et al. [26], where subcutaneous metastases were adverse prognosticator of survival, as well as studies of Paesmans et al. [33] and Hikisch et al. [42], who both found an adverse influence of the presence of skin metastases on survival. It is, therefore, likely that no rational application of the data from the literature could successfully lead to recommendations for future studies regarding the metastatic site as a potential prognosticator of survival in this disease. The interpretation of the findings of some of the studies discussed above must be approached with some caution, because it is a well-recognized fact that the small numbers of patients in some of the studies may

severely distort the overall findings leading to falsepositive results. To combat this, the ratio of the number of patients suffering endpoints to the number of potential prognostic factors was proposed as an indication of the ‘‘effectiveness’’ of the analysis. It was suggested that this ratio, although rather arbitrary, should, for example, be >10 [44]. Our study fits well within this framework, with this ratio being 24. In conclusion, our study showed that gender, performance status, pretreatment weight loss, number of metastatic sites as well as the presence of liver and brain metastases are independent prognosticators of survival in patients with stage IV NSCLC treated with CHT. These results add to the ongoing controversy on the identification of potential prognostic factors in this disease and their use in its treatment.

References 1. Ihde DC (1992) Chemotherapy of lung cancer. N Engl J Med 327:1434–1441 2. Souquet PJ, Chauvin F, Boissel JP, et al (1993) Polychemotherapy in advanced non-small cell lung cancer: a meta analysis. Lancet 342:19–21 3. Grilli R, Oxman AD, Julian JA (1993) Chemotherapy for advanced non-small cell lung cancer: how much benefit is enough? J Clin Oncol 11:1866–1872 4. Non-Small Cell Lung Cancer Cooperative Group (1995) Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomized clinical trials. BMJ 311:899–909 5. Marino P, Preatoni A, Cantoni G, Buccheri G (1995) Singleagent chemotherapy versus combination chemotherapy in advanced non-small cell lung cancer: a quality and meta-analysis study. Lung Cancer 13:1–12 6. Lilenbaum RC, Langenberg P, Dickersin K (1998) Single agent versus combination chemotherapy in patients with advanced nonsmall cell lung cancer. A meta-analysis of response, toxicity, and survival. Cancer 82:116–126 7. Depierre A, Chastang C, Quoix E, et al (1994) Vinorelbine versus vinorelbine plus cisplatin in advanced non-small cell lung cancer: a randomized trial. Ann Oncol 5:37–42 8. Wozniak AJ, Crowley JJ, Balcerzak SP, Weiss GR, Spiridonidis CH, Albain KS, Kelly K, Taylor SA, Gandara DR, Livingston RB (1998) Randomized trial comparing cisplatin with cisplatin plus vinorelbine in the treatment of advanced nonsmall cell lung cancer: a Southwest Onocology Group Study. J Clin Oncol 16:2459–2465 9. Cardenal F, Lopez-Cabreirizo MP, Anton A, Alberola V, Massuti B, Carrato A, Barneto I, Lomas M, Garcia M, Lianes P, Montalar J, Vadell C, Gonzalez-Larriba JL, Nguyen B, Artal A, Rosel R (1999) Randomized phase III study of gemcitabine-cisplatin versus etoposide-cisplatin in the treatment of locally advanced or metastatic non-small cell lung cancer. J Clin Oncol 17:12–18 10. Sandler AB, Nemunaitis J, Denham C, von Pawel J, Cormier Y, Gatzemeier U, Mattson K, Manegold C, Palmer MC, Gregor A, Nguyen B, Niyikiza C, Einhorn LH (2000) Phase III trial of gemcitabine plus cisplatin versus cisplatin alone in patients with locally advanced or metastatic non-small cell lung cancer. J Clin Oncol 19:122–130 11. Bonomi P, Kim K, Fairclough D, et al (2000) Comparison of survival and quality of life in advanced non-small cell lung cancer patients treated with two dose levels of paclitaxel combined with cisplatin versus etoposide with cisplatin: results of an Eastern Cooperative Oncology Group trial. J Clin Oncol 18:623–631

122 12. Giaccone G, Splinter TA, Debruyne C, Kho GS, Lianes P, van Zandwijk N, Pennucc MC, Scagliotti G, van Meerbeeck J, van Hoesel Q, Curran D, Sahmoud T, Postmus PE (1998) Randomized study of paclitaxel-cisplatin versus cisplatin-teniposide in patients with advanced non-small cell lung cancer. The European Organization for Research and Treatment of Cancer Lung Cancer Cooperative Group. J Clin Oncol 16:2133–2141 13. Gatzemeier U, von Pawel J, Gottfried M, ten Velde GP, Mattson K, DeMarinis F, Harper P, Salvati F, Robinet G, Lucenti A, Bogaerts J, Gallant G (2000) Phase III comparative study of high-dose cisplatin versus a combination of paclitaxel and cisplatin in patients with advanced non-small cell lung cancer. J Clin Oncol 18:3390–3399 14. Kelly K, Crowley J, Bunn PA Jr, Presant CA, Grevsted PK, Moinpour CM, Ramsey SD, Wozniak AJ, Weiss GR, Moore DF, Israel VK, Livingston RB, Gandara DR (2001) Randomized phase III trial of paclitaxel plus carboplatin versus vinorelbine plus cisplatin in the treatment of patients with advanced non-small cell lung cancer: a Southwest Oncology Group trial. J Clin Oncol 19:3210–3218 15. Kosmidis P, Mylonakis N, Nicolaides C, Kalophonos C, Samantas E, Boukovinas J, Fountzilas G, Skarlos D, Economopolous T, Tsavaridis D, Papakostas P, Bacoyiannis C, Dimopolous M (2002) Paclitaxel plus carboplatin versus gemcitabine plus paclitaxel in advanced non-small cell lung cancer: a phase III randomized trial. J Clin Oncol 20:3578–3585 16. Schiller JH, Harrington D, Belani C, Langer C, Sandler A, Krook J, Zhu J, Johnson DH (2002) Comparison of four chemotherapy regimens in advanced non-small cell lung cancer (NSCLC). N Engl J Med 346:92–98 17. Jeremic B, Shibamoto Y, Acimovic LJ, Milisavljevic S, Milicic B, Nikolic N (1997) Prolonged administration of oral etoposide alone or with intravenous carboplatin in stage IV non-small cell lung cancer: a randomized trial. Lung Cancer 18:179–188 18. Jeremic B, Shibamoto Y, Milicic B, et al (1999) Prolonged oral versus high-dose intravenous etoposide in combination with carboplatin for stage IV non-small cell lung cancer. A randomized trial. Lung Cancer 25:207–214 19. Hainsworth JD, Johnson DH, Frazier SR, et al (1989) Chronic daily administration of oral etoposide: a phase I trial. J Clin Oncol 7:396–401 20. Ederer F, Mersheimer WL (1962) Sex differences in the survival of lung cancer patients. Cancer 15:425–432 21. Connelly RR, Cutler SJ, Baylis P (1966) End results in cancer of the lung: comparison of male and female patients. J Natl Cancer Inst 36:277–287 22. Watson WL, Schottenfeld D (1968) Survival in cancer of the bronchus and lung, 1949–1962; comparison of men and women patients. Dis Chest 53:65–72 23. Glatstein E, Makuch RW (1984) Illusion and reality: practical pitfalls in interpreting clinical trials. J Clin Oncol 2:488–497 24. Simon R (1984) Importance of prognostic factors in cancer clinical trials. Cancer Treat Rep 68:185–192 25. Miller TP, Chen TT, Coltman CA Jr, et al (1986) Effect of alternating combination chemotherapy on survival of ambulatory patients with metastatic large-cell and adenocarcinoma of the lung. A Southwest Oncology Group Study. J Clin Oncol 4:502–508 26. Finkelstein DM, Ettinger DS, Ruckdeshhel JC (1986) Longterm survivors in metastatic non-small cell lung cancer: an Eastern Cooperative Oncology Group study. J Clin Oncol 4:702–709 27. Lanzotti VJ, Thomas DR, Boyle LE, Smith TL, Gehan EA, Samuels ML (1977) Survival with inoperable lung cancer. An integration of prognostic variables based on simple clinical criteria. Cancer 39:303–313

28. Stanley KE (1980) Prognostic factors for survival in patients with inoperable lung cancer. J Natl Cancer Inst 65:25–32 29. Sakurai M, Shinkai T, Eguchi K, et al (1987) Prognostic factors in non-small cell lung cancer: multioregressional analysis in the National Cancer Center Hospital (Japan). J Cancer Res Clin Oncol 113:563–566 30. Einhorn LH, Loehrer PJ, Williams SD, et al (1986) Random prospective study of vindesine versus vindesine plus high-dose cisplatin versus vindesine plus cisplatin plus mitomycin C in advanced non-small cell lung cancer. J Clin Oncol 4: 1037–1043 31. O’Connell JP, Kris MG, Gralla RJ, et al (1986) Frequency and prognostic importance of pretreatment clinical characteristics in patients with advanced non-small cell lung cancer treated with combination chemotherapy. J Clin Oncol 4:1604–1614 32. Sorensen JB, Badsberg JH, Olsen J (1989) Prognostic factors in inoperable adenocarcinoma of the lung: a multivariate regression analysis of 259 patients. Cancer Res 49:5748–5754 33. Paesmens M, Sculier JP, Libert P, et al (1995) Prognostic factors for survival in advanced non-small cell lung cancer: univariate and multivariate analyses including recursive partitioning and amalgamation algorithms in 1,052 patients. The European Lung Cancer Working Party. J Clin Oncol 13:1221–1230 34. Kawahara M, Furuse K, Kodama N, et al (1991) A randomized study of cisplatin versus cisplatin plus vindesine for non-small cell lung carcinoma. Cancer 68:714–719 35. Rapp E, Pater JL, Willan A, et al (1988) Chemotherapy can prolong survival in patients with advanced non-small cell lung cancer—Report of a Canadian multicenter randomized trial. J Clin Oncol 6:633–641 36. Shinkai T, Eguchi K, Sasaki Y, et al (1992) A prognostic-factor risk index in advanced non-small cell lung cancer treated with cisplatin-containing combination chemotherapy. Cancer Chemother Pharmacol 30:1–6 37. Albain KS, Crowley JJ, LeBlanc M, Livingston RB (1991) Survival determinants in extensive-stage non-small cell lung cancer: the Southwest Oncology Group experience. J Clin Oncol 9:1618–1626 38. Ewans WK, Nixon DW, Daly JM, et al (1987) A randomized study of oral nutritional support versus ad lib nutritional intake during chemotherapy for advanced colorectal and non-small cell lung cancer. J Clin Oncol 5:113–124 39. Sculier JP, Paesmens M, Libert P, et al (1994) Long-term survival after chemotherapy containing platinum derivatives in patients with advanced unresectable non-small cell lung cancer. Eur J Cancer 30A:1342–1347 40. Fukuoka M, Masuda N, Furuse K, et al (1991) A randomized trial in inoperable non-small cell lung cancer: vindesine and cisplatin versus mitomycin, vindesine, and cisplatin versus etoposide and cisplatin alternating with vindesine and mitomycin. J Clin Oncol 9:606–613 41. Ferguson MK, Skosey C, Hoffman PC, Golomb HM (1990) Sex-associated differences in presentation and survival in patients with lung cancer. J Clin Oncol 8:1402–1407 42. Hickish TF, Smith IE, O’Brien MER, Ashley S, Middleton G (1998) Clinical benefit from palliative chemotherapy in nonsmall cell lung cancer extends to the elderly and those with poor prognostic factors. Br J Cancer 78:28–33 43. Green N, Kurohara S, George F (1971) Cancer of the lung, an in-depth analysis of prognostic factors. Cancer 28:1229–1233 44. Harrell FE, Lee KL, Matchar DB, Reichert TA (1985) Regression models for prognostic prediction: advantages, problems, and suggested solutions. Cancer Treat Rep 69:1071–1077