Today, lung cancer represents a major public health concern worldwide, accounting for about 12% of all new cancers in both sexes [1]. Small cell lung cancer (SCLC), accounting for 20% to 25% of all lung cancers, is highly sensitive to chemotherapy and radiotherapy. Optimal treatment yields a high response rate (60%-90%), but most patients relapse and will eventually die of chemotherapy-resistant disease [2]. SCLC, however, is potentially curable, with a 5-year survival rate of 7% to 15% in limited-stage disease (LD) [3]. The recognition of prognostic factors in SCLC has several aims including the following: 1. Individual prognostic counseling 2. Selecting treatment, when therapeutic options depend on the baseline clinical characteristics of the subject 3. Adjusting for inhomogeneities when comparing groups of patients from different locations and studies 4. Defining the eligibility criteria for new clinical trials and stratifying patients by risk subgroups 5. Understanding certain factors that may provide insights into the disease process and provide direction for further studies Although some of these reasons may appear rather theoretic, they do have practical relevance. As for many other solid tumors, the variability in prognosis among SCLC patients is more substantial than improvements in prognosis due to therapy. The failure to appreciate adequately the importance of prognostic factors may contribute to the design of inefficient studies, the erroneous interpretation of results, and the development of an inconsistent literature. A recent development of the research on prognostic factors has been made possible by the widespread use of data-recording systems based on microcomputers. Single institutions are now able to analyze their own data or to cumulate them into large multi-institutional files by exchanging magnetic support media. As a result, the number of studies dealing with newer and more traditional prognostic factors is proliferating. A MEDLINE search using the terms prognostic factor and cancer produced 2393 articles, according to a recent editorial [4]. It is unfortunate that the interpretation of this vast and ever-growing literature is not easy [5]. Apart from the magnitude itself of the relevant literature, there is a remarkable inhomogeneity among studies. Major differences concern the following: 1. Study populations (eg, LD and extensive disease [ED]) 2. Diagnostic criteria and treatment modalities 3. Statistical analysis (eg, the univariate survival estimate of Kaplan-Meyer, the multivariate Cox logistic regression, and the recursive partitioning and amalgamation algorithms [RECPAM]) 4. The mix of variables that are taken into consideration 5. The inclusion of post-treatment factors such as response to treatment 6. The endpoints themselves (eg, the entire survival curve or the survival at particular times, usually survival rates at 2 or 5 years) Traditionally, the anatomic extent of disease, performance status (PS), and weight loss (WL) have been used to predict the outcome of patients with SCLC [5-10]; however, simple biochemical tests and serum tumor markers are also predictive of survival. A multiple regression analysis identified PS, disease extent, AP, sodium, and albumin as contributing independently to survival. Using these parameters, three prognostic groupings could be defined. The patients in the "best" prognostic group were found to have higher response to treatment and longer survival times [15]. In 407 SCLC patients, 61 pretreatment variables were evaluated in a Cox multiple regression analysis by Cerny and coworkers [ 16]. LDH (P = 0.001), tumor stage (P = 0.0001), serum sodium (P = 0.0009), pretreatment KPS (P = 0.0121), AP (P = 0.0186), and serum bicarbonate (P = 0.0321), but not albumin, were significant prognostic factors [16]. The study reported by the Subcommittee for the Management of Lung Cancer (United Kingdom Coordinating Committee on Cancer Research) merits mentioning [66]. This study collected information from 10 different British SCLC studies on 17 laboratory tests and nearly 4000 patients and concluded that the most useful laboratory tests are AP, glutamic oxalacetic transaminase, LDH, and sodium [66]. In the opinion of the subcommittee, these parameters should be measured in any future SCLC study [66]. Recently, hemoglobin, leukocyte, and platelet counts were found to be independent prognostic factors in 436 patients included in a prospective multicenter SCLC study with a minimum of 5-year follow-up [36].
