Small-cell lung carcinoma (SCLC) accounts for 20% to 25% of all new cases of lung cancer in the United States. It is estimated that approximately 42,000 new cases will occur in the United States in 1995 [1,2]. Of the various histologic types of lung cancer, small-cell is the most sensitive to chemotherapy and radiotherapy, yet overall outcome is poor, with only 5% to 10% of patients surviving 5 years from diagnosis. Thus, SCLC has a generally grim prognosis. However, an orderly approach to diagnosis, staging, and treatment based on knowledge of the clinical behavior of the disease allows for the selection of the best curative and palliative therapy for individual patients.
In this review, the epidemiology, etiology, pathology, and natural history of the disease are discussed. In addition, current clinical management is examined.
Cigarette smoking remains the major cause of SCLC, with more than 90% of cases in men attributable to tobacco use. The recent increase in SCLC in women is associated with a rise in cigarette smoking among women over the past two to three decades; indeed, SCLC seems to be the most rapidly increasing type of lung cancer in women .
There is a dose-response relationship between the risk of death from SCLC and the total number of cigarettes smoked such that the risk is increased 60- to 70-fold for the smoker with 40 pack-years' exposure compared with the nonsmoker. Conversely, the chance of developing SCLC decreases with cessation of smoking but, especially in the heavy smoker (20 pack-years or more), may never return to the level of risk of the nonsmoker. Among long-term survivors of SCLC, smoking cessation decreases the risk of a second primary (non-small-cell) lung cancer, whereas for patients who continue to smoke, the risk of a second malignancy is increased about 32-fold .
Exposure to industrial or environmental toxins, such as asbestos and radon, in combination with tobacco use synergistically increases the risk of developing SCLC and other lung cancers. Exposure to halogenated ethers also has been associated with an increased risk of SCLC.
In both genders, the majority of SCLC cases occur in the age range of 35 to 75 years, with a peak at age 55 to 65 years.
Lung cancer is the final step in a series of morphologic and structural changes that take several years to occur. Multiple genetic changes have been elucidated and likely play a role in field cancerization, through which all the aerodigestive tract epithelium is mutagenized by tobacco and other carcinogens. Genetic lesions may prove useful in the early detection, prevention, and treatment of lung carcinoma in the future.
Tumor-supressor gene products are negative growth regulators in the cell cycle. As opposed to classical oncogenes, tumor-suppressor genes are oncogenic through loss of function rather than activation. The first evidence for their existence came from somatic-cell fusion studies that demonstrated the recessive nature of the tumorigenic phenotype. Many years following that first observation, restriction fragment-length polymorphism (RFLP) analysis of tumor and normal tissue from children with retinoblastoma showing loss of heterozygosity (loss of one allele) in tumor tissue provided strong support for loss of putative tumor-suppressor elements as oncogenic. Further studies with retinoblastoma demonstrated that this loss of heterozygosity was correlated with loss of function of both alleles of a single gene, the retinoblastoma (Rb) gene on chromosome 13. Application of this methodology to the common solid tumors of adulthood revealed frequent loss of heterozygosity in diverse chromosomal regions known or postulated to be loci for tumor-suppressor genes. This phenomenon is especially common in SCLC.
The Rb gene product is a nuclear phosphoprotein involved in cell cycle regulation. Frequent cytogenetic alterations at 13q14, the locus for this gene, in SCLC cell lines and tumors were the catalyst for further study of this area. Analysis of mRNA expression shows that 60% of SCLC cell lines have undetectable transcripts and that the remaining 40% have an abnormal gene product. In addition, evaluation of DNA from 26 SCLC tumors and normal tissues indicated that six of six patients whose normal DNA was heterozygous for an Rb RFLP had lost one of the two alleles in their tumor DNA . These data support the concept of inactivation of the Rb gene as a key event in the pathogenesis of SCLC.
Essentially all SCLC tumors and cell lines have deletions of the short arm of chromosome 3 at 3p21 distal-p22 [6-10]. Although the precise identity of this gene awaits further investigation, several candidate genes have been studied. These include the genes encoding protein-tyrosine phosphatase-gamma, beta-retinoic acid receptor, and a serine/threonine protein kinase (c-raf-1) . However, there is no direct evidence supporting the role of any of these as tumor suppressors in SCLC. Because many other solid tumors also contain deletions in this area of the genome, it remains a fertile area for research.
Mutations in the p53 tumor-suppressor gene on chromosome 17p are the most commonly identified genetic alteration in human cancers and have been documented in 60% to 100% of SCLC cell lines and in 77% of tumors [12-14]. The normal product of the p53 locus acts to suppress and control cell division, and its loss may favor unregulated growth.
Proto-oncogenes play an important role in normal cell growth. Abnormal expression of these genes by point mutations, translocation, or DNA amplification may confer tumorigenicity. Oncogene products are positive effectors of transformation, superimposing their activity on the cell with a gain in function.
Overall, amplification of the myc family of oncogenes (c-myc, N-myc, and L-myc) has been identified in 36 (18%) of 200 tumors and in 38 (31%) of 122 tumor-cell lines from patients with SCLC (see reference 12). Amplification of Myc family DNA is more frequent in tumor-cell lines than in tumor specimens and is also more frequent in tumors from previously treated patients. In addition, some studies have suggested that patients with SC/LC histology possessing c-myc amplification have a shorter survival, reflecting a more agressive and treatment-resistant tumor [15-17]. However, because these studies used tumor samples from treated patients, the significance of this finding is not clear.
SCLC produces a variety of peptide hormones that act as mitogens and growth factors, eg, insulin-like growth factor-I (IGF-I) and gastrin-releasing peptide . These peptides participate in an autocrine loop resulting in growth of the tumor mass and sustenance of hormone production. This autocrine pathway provides an opportunity for therapy, and in fact, a phase II trial with a murine monoclonal antibody against gastrin-releasing peptide for recurrent SCLC resulted in 1 patient with a complete response and 4 with stable disease among 12 patients .
IGF-I is present in 95% of SCLC tumors and cell lines, and there is a suggestion that it is also an autocrine growth factor. Indeed, somatostatin(Drug information on somatostatin) analogs inhibit the production of IGF-I and are active against SCLC xenografts in rats .
The development of drug resistance is one of the most intriguing phenomena in the biology of SCLC. Several events at the cellular level have been described in cell lines made resistant by repeated exposure to cytotoxic agents in vitro.
The best established mechanism for multidrug resistance in cancer cells is overexpression of the multidrug-resistance (MDR-1) gene, encoding the multidrug efflux pump termed p-glycoprotein (p for “permeability”). Multidrug resistance to vinca alkaloids, anthracyclines, and epipodophyllotoxins results from the action of the p-glycoprotein in active transport of drug across the cytoplasmic membrane.
The role of MDR-1 gene expression in SCLC is controversial. No correlation between gene expression and previous exposure to chemotherapy, response to chemotherapy, or in vitro drug sensitivity was found among a panel of six SCLC tumor specimens and 23 SCLC cell lines . However, Poupon et al used tumor samples transplanted into nude mice and correlated responses to treatment in the patients and in the xenografts with MDR-1 mRNA levels in the xenografts. Two patients who achieved complete responses and long-term survival had a complete xenograft response and no detectable level of MDR-1 mRNA. Three patients who had recurrences after complete responses had MDR-1 transcript detected in the pretreatment specimen, and one patient who at the time of recurrence had biopsies before and after chemotherapy had no detectable MDR-1 mRNA in the first sample but transcripts clearly present in the posttreatment specimen . Although these data are intriguing, the small number of patients studied calls for caution in interpretation.
Given the broad spectrum of drug resistance that characterizes recurrent SCLC, it seems probable that multiple mechanisms are operative simultaneously. Another transporter gene distantly related to MDR-1 and in the same superfamily of ATP-binding cassette transmembrane transporters encodes a protein termed the multidrug-related protein (MRP). Overexpression of the MRP gene has been found in several multidrug-resistant SCLC cell lines selected in vitro [23,24]. A study of 12 fresh tumor samples revealed low expression of the MRP gene (relative to the level of sensitive SCLC lines) in 4, increased expression (1.5- to 2-fold) in 5, and 3 with no detectable MRP expression . Other investigators analyzed 12 unselected SCLC cell lines and were unable to correlate MRP gene expression with chemosensitivity to doxorubicin(Drug information on doxorubicin) (Adriamycin, Rubex), etoposide(Drug information on etoposide) (VePesid), and cisplatin(Drug information on cisplatin) (Platinol).
In multidrug-resistant cells that do not overexpress MDR-1 or MRP, changes in activity or in the amount of topoisomerase I or II may be the basis for the resistance to epipodophyllotoxins and anthracyclines or to camptothecin and its analogs, respectively. Topoisomerases modify the topologic structure of DNA without changing the nucleotide sequence, cutting one (topoisomerase I) or both (topoisomerase II) strands of DNA and allowing strand passage through the transient nick, followed by religation. Topoisomerase-targeted drugs stabilize the usually transient DNA-enzyme complex, the so-called cleavable complex, and in so doing produce aberrations in DNA function that can initiate a process that culminates in cell death. Lower levels of expression or mutations in topoisomerases I and II in cell lines selected in vitro are associated with resistance to drugs that target the respective enzymes.
In addition, topoisomerase II expression has been associated with a multidrug-sensitive phenotype in lung cancer. Levels of topoisomerase II gene expression and sensitivity to doxorubicin, etoposide, teniposide(Drug information on teniposide) (Vumon), cisplatin, camptothecin, and fluorouracil(Drug information on fluorouracil) were directly correlated in a study of eight human lung cancer cell lines . Cell lines expressing this multidrug-sensitivity phenotype were sensitive not only to drugs that directly inhibit topoisomerase II but also to drugs with other clearly defined mechanisms of action, such as camptothecin and cisplatin. Giaccone et al  hypothesized that topoisomerase II is a key participant in a final common pathway of cell death initiated by various macromolecular insults.
With the advent of the 1967 World Health Organization (WHO) classification system for SCLC, several subtypes were suggested based on morphologic features (Table 1). However, subtyping SCLC into categories, such as lymphocyte-like, fusiform, and polygonal, was unreliable because judgments varied among observers from case to case and even within individual cases. In 1977, the second WHO update simplified the SCLC subtyping and designated three categories: oat-cell carcinoma, intermediate small-cell carcinoma, and combined small-cell carcinoma. However, this revised system did not allow for the previously recognized morphologic variant subtype small-cell/large-cell (SC/LC) carcinoma (Table 1) . In some studies, this SC/LC subtype correlated with an unfavorable prognosis and appeared to have greater resistance to radiation and chemotherapy. In addition, the second WHO update separated the intermediate-cell subtype from the small-cell subtype, an artificial separation not justified by differences in clinical behavior. As a result, the International Association for the Study of Lung Cancer (IASLC) pathology panel redefined the subtypes of SCLC in an attempt to make them more clinically relevant (Table 1).
|World Health Organization I (1967)|
Lymphocyte-like ("oat cell")
|World Health Organization II (1977)|
|International Association for the Study of Lung Cancer (1994)|
|Reprinted, with permission, from Khouri IF, Glisson BS: Small-cell lung cancer, in Pazdur R (ed): Medical Oncology: A Comprehensive Review, p 120. Huntington, NY, PRR Inc, 1993.|
Typical SCLC is characterized by hyperchromatic nuclei with a very high nucleus to cytoplasm ratio; a diffuse, finely stippled pattern of chromatin; and indistinct nucleoli. The cells are small, approximately two to three times the size of a lymphocyte. The intermediate-cell subtype shows fusiform or polygonal cells with granular chromatin, inconspicuous nucleoli, and modest amounts of cytoplasm.
Variant-morphology SCLC is marked by distinct clusters of large cells exhibiting vesicular nuclei with finely dispersed chromatin and prominent nucleoli integrally admixed in what would otherwise be a typical SCLC. Because mixed histology in SCLC is not uncommon, especially at autopsy (30%), and in view of in vitro transformation of SCLC to a larger cell type with loss of neuroendocrine features, it has been proposed that initiation of pulmonary carcinogenesis occurs in a pluripotent stem cell capable of differentiation along different pathways.
SCLC expresses several markers of neuroendocrine differentiation, such as chromogranin and gastrin-releasing peptide. A new nomenclature encompassing carcinoids, atypical carcinoids, large-cell neuroendocrine carcinoma, and SCLC and forming a spectrum of neuroendocrine neoplasia has been proposed, with the two ends of this biologic and clinical spectrum occupied by the indolent bronchial carcinoid tumor and the aggressive small-cell anaplastic carcinoma . In addition to the spectrum of pathologic and biologic features, these tumors differ in their sensitivity to cytotoxic agents with response rates increasing as the histology becomes more anaplastic.
No molecular targets specific for SCLC have been completely defined, preventing the use of molecular biology techniques in the diagnosis of SCLC. Deletions in the short arm of chromosome 3, for example, are highly prevalent in SCLC but can be seen in non-small-cell lung cancer and in many other cancers as well [31-33].
The natural history of SCLC is characterized by a relentlessly progressive clinical course and by the early development of metastatic disease. If untreated, this disease is rapidly fatal, with a median survival of 5 to 7 weeks in patients with clinically apparent metastatic disease and 12 weeks in patients with regional thoracic involvement only . Based on the results of an early study, even patients who have seemingly localized disease are presumed to have occult metastatic involvement. In that study, 19 patients with SCLC underwent purportedly curative surgical resection but died within 30 days of non-cancer-related causes. Of those 19 patients, 70% had distant metastatic disease at autopsy . It is not surprising, therefore, that localized forms of treatment used alone fail in the majority of patients. Although radiotherapy was shown to be superior to surgery alone in prolonging median survival in a prospective randomized trial conducted in patients with operable local-regional disease, less than 5% of patients in the study were alive 5 years after treatment with either modality .
In 1969, the Veterans Administration Lung Cancer Study Group (VALCSG) reported that three courses of cyclophosphamide(Drug information on cyclophosphamide) more than doubled the median survival of patients with advanced SCLC . Subsequently, over the past two decades of clinical research, it has become clear that SCLC is fundamentally a systemic, not a localized, disorder and that it is the histologic type of lung cancer most responsive to cytotoxic agents.
Compared with patients who have other types of lung cancer, those with SCLC report a shorter duration of symptoms before diagnosis, usually less than 3 months. Because SCLC lesions are usually centrally located, they can induce cough, wheezing, stridor, deep chest pain, hemoptysis, and dyspnea caused by airway obstruction, with or without postobstructive pneumonitis.
Mediastinal involvement, either by direct extension or by lymphatic metastases, is a hallmark of SCLC and can result in a variety of complications. Nerve entrapment can lead to recurrent laryngeal nerve paralysis and hoarseness. Involvement of the phrenic nerve can cause paralysis and elevation of the hemidiaphragm, with resulting dyspnea. Tumor compression of the esophagus can lead to dysphagia. Frequently, a right-sided hilar mass or tumor in the right mediastinal lymph nodes compresses the thin-walled, low-pressure system of the superior vena cava, causing superior vena cava syndrome .
SCLC is particularly notable for producing a variety of paraneoplastic syndromes that can be the first indication of the tumor's presence. Classic examples include hyponatremia caused by production of arginine vasopressin(Drug information on vasopressin) and possibly atrial natriuretic factor  and Cushing's syndrome with or without hyperpigmentation caused by excessive production of the precursor peptides of ACTH (pro-opiomelanocortin and pro-ACTH) . Also, the development of auto-antibodies to normal neural antigens in patients with SCLC can lead to Eaton-Lambert syndrome, retinal blindness, subacute sensory neuropathy, and subacute cerebellar degeneration. Recently, cognitive dysfunction has been documented in patients with completely responding SCLC prior to elective brain irradiation , and it is hypothesized that this is yet another manifestation of an autoimmune paraneoplastic syndrome.
One of the most distressing syndromes is weight loss and anorexia, which occurs in nearly one third of SCLC patients and for which no causative mechanism is currently known. Experimental models, however, suggest that the actions of various cytokines, including interleukin-1, tumor necrosis factor, and interleukin-6, may be involved in the pathogenesis of this syndrome.
The development of a common staging system for malignant diseases has enabled physicians to accurately assess and compare the results of therapy. It has also permitted the elaboration of prognostic factors that can improve the design of future treatments. The American Joint Committee on Cancer has developed a staging system for lung cancer based on the size and extent of the primary tumor (T), the degree of regional lymph-node involvement (N), and the presence or absence of distant metastases (M) .
Although this TNM staging system has been shown to be of critical importance in staging non-small-cell lung cancer, TNM stage has been prognostically significant in SCLC. In a series of patients studied in the prechemotherapy era, patients who had no extrathoracic metastases and who were selected to undergo curative resections based on T and N classification survived no longer than did patients whose tumors were deemed unresectable based on the extent of regional disease . Based in part on these data, surgical treatment of SCLC was abandoned. It should be noted, however, that there were very few patients with stage I tumors in that series.
An alternative staging system more clearly prognostic for patients with SCLC was proposed by the VALCSG. This classification divided SCLC into limited disease (LD) and extensive disease (ED), a division based on the clear survival differences observed in earlier studies between patients with regional (or limited) disease and those with distant (or extensive) spread . LD as originally defined included tumors whose total known extent could be covered in one radiotherapy port. This original classification of LD included involvement of ipsilateral and contralateral hilar, mediastinal, and supraclavicular nodes as well as ipsilateral pleural effusion. However, as therapy for LD has evolved and the value of chest radiation has become apparent for a subset of patients, a refined definition of LD excluding contralateral hilar and contralateral supraclavicular nodal involvement as well as pleural effusion is frequently used. This definition identifies patients who are likely to have disease amenable to a combined chemotherapy and radiation (chemoradiation) approach with curative intent and to have disease that will be covered in a tolerable port size. ED denotes tumor beyond these limits and usually accounts for 60% to 70% of patients with SCLC. Long-term follow-up of patients treated with chemotherapy has revealed that survival of more than 3 years occurs almost exclusively in those with LD and that long-term survival in patients with ED remains uncommon . These data indicate that disease classification by the VALCSG staging system is an independent prognostic variable and support its continued use.
The role of TNM classification in identifying patients with SCLC who might benefit from a surgical approach has been reexamined in the past decade. While data from a recent randomized trial show no benefit for adjuvant resection for stage II to III disease, it is fairly clear that patients who present with SCLC at clinical stage I (solitary pulmonary nodule) do very well after initial resection and adjuvant chemotherapy [44,44a]. Thus, TNM staging can be used to define the subset of patients with LD who could benefit from an approach including resection.
Multivariate analyses of large databases have determined that disease extension and performance status are the most important prognostic factors. Female sex correlated with better response rates and survival in one study by the Cancer and Leukemia Group B . Advanced age (older than 60 years), supraclavicular nodal involvement in LD patients, and a large number of metastatic sites also were predictive of a poor outcome. Other poor prognostic signs are low levels of hemoglobin, platelets, sodium, albumin, uric acid, and bicarbonate; elevated levels of leukocytes, lactate dehydrogenase, carcinoembryonic antigens; weight loss; slow response to therapy, and possibly failure to achieve a complete response .
Although staging is important in assigning prognosis and in locating sites of disease to determine response to therapy, its most critical role is in identifying LD patients who will benefit from an aggressive combined-modality approach with chemoradiation. Since the most important risk factor for treatment-related death appears to be initial performance status, evaluation of the physiologic condition of the patient is required. The initial survey should begin with a complete history and physical examination, chest roentgenography, and hematologic and clinical profile. The usual central location of disease, extensive mediastinal involvement, and postobstructive changes frequently make it difficult to define measurable disease by plain chest films alone. For this reason, computed tomography (CT) scans can aid greatly in the primary staging evaluation, in planning thoracic radiotherapy, and also in evaluating the mediastinal and hilar nodes in patients in whom surgical removal of the primary tumor is being considered.
Involvement of intra-abdominal organs is noted in 30% of patients undergoing initial staging for SCLC , with liver, adrenal gland, pancreas, and intra-abdominal (mesenteric and retroperitoneal) lymph nodes representing the most frequent metastatic sites. CT is probably the best method for defining such intra-abdominal organ involvement. Since normal results of liver function tests, such as levels of serum glutamic-oxaloacetic transaminase (SGOT), lactate dehydrogenase, and alkaline phosphatase, do not exclude liver metastases, a CT scan of the abdomen should be a routine component of the staging evaluation.
Radionuclide bone scans remain the most valuable screening test for detecting early bone metastases, which occur in nearly 30% of patients who present with SCLC. Screening bone scans in asymptomatic patients can yield positive results even when bone pain is absent and alkaline phosphatase levels are normal. Only a small percentage of patients with abnormal findings on bone scans will have positive findings on skeletal radiographs, but radiographs may be useful in excluding benign lesions corresponding to abnormal areas on bone scans.
Bone marrow involvement is present at diagnosis in 20% to 25% of patients with SCLC, and the majority of those have normal peripheral blood findings . Therefore, in the absence of other documented sites of extensive disease, a routine unilateral bone marrow examination is warranted in SCLC. However, bilateral aspirations and biopsies, which would increase the diagnostic yield by 30% but would affect the staging status (and thus alter therapy) in only 5% of cases, are not recommended routinely . The results of bone scans and marrow examinations do not necessarily correlate; rather, the two tests are complementary, each reflecting a separate pattern of disease spread.
The brain is also commonly involved at presentation in SCLC patients. About 10% of patients have symptoms related to cerebral metastases, and another 5% have asymptomatic involvement . Thus, magnetic resonance imaging or CT of the brain is a routine part of complete staging. Carcinomatous leptomeningitis is the second most common form of central nervous system involvement in SCLC but is rarely a problem at presentation (less than 0.5% of patients). For this reason, routine evaluation of cerebrospinal fluid is not indicated in the absence of specific symptoms.
Serum markers as diagnostic tools or as prognostic factors have low positive and negative predictive values. Pretreatment levels of neuron-specific enolase as predictors of disease-free survival have a sensitivity of 65% to 79% and a specificity of 82% to 86%; the overlap between the values for LD and ED prevents the use of this marker as an accurate determinant of disease stage. Serum levels of neuron-specific enolase and carcinoembryonic antigen may correlate with tumor bulk and response to chemotherapy [51,52].
Expansion of the initial staging evaluation depends on the clinical circumstances. In particular, when patients are not being treated as part of a clinical research protocol, staging can be clinically directed toward obvious sites of metastasis. In this setting, the identification of asymptomatic metastatic disease is unlikely to affect management and outcome. A possible exception is the case of occult brain lesions, for which early radiation may reduce later morbidity from symptomatic brain metastases.
Approximately two thirds of patients with SCLC present with overt evidence of tumor dissemination beyond the thorax, and most of those with seemingly localized disease are presumed to have occult metastatic involvement . This early metastatic involvement and frequent regional nodal spread support the idea that SCLC is a systemic disease at the time of initial clinical presentation. The systemic nature of the disease and its marked sensitivity to cytotoxic agents have resulted in the establishment of chemotherapy as the foundation for treatment of SCLC.
Following the initial VALCSG report showing that treatment with cyclophosphamide (Cytoxan, Neosar) resulted in prolonged survival for SCLC patients, many clinical trials were undertaken in the 1970s to study the efficacy of chemotherapy in this disease. Over the past two decades, many single agents, including cyclophosphamide and ifosfamide(Drug information on ifosfamide) (Ifex), doxorubicin and epirubicin(Drug information on epirubicin) (Farmorubicin), etoposide and teniposide, cisplatin and carboplatin (Paraplatin), vincristine (Oncovin), methotrexate(Drug information on methotrexate), and the nitrosoureas have shown activity against SCLC, with response rates ranging from 15% to 50% [53-57]. Most recently, paclitaxel(Drug information on paclitaxel) (Taxol) and the topoisomerase I inhibitors topotecan(Drug information on topotecan) and irinotecan(Drug information on irinotecan) have also been found active [58-60].
The activity of the different agents in SCLC appears to be highly dependent on prior therapy. For instance, the early experience with short-course intravenous etoposide showed a response rate of only 9% to 15% when used in patients with treatment-refractory disease ; yet when etoposide was administered to previously untreated patients, 40% to 60% had an objective response [54,56]. This has led to the suggestion that phase II trials should be performed in previously untreated patients with ED. While this approach remains somewhat controversial , randomized studies have demonstrated its feasibility and safety provided that effective salvage chemotherapy is administered promptly for progressive disease [62,63].
The involvement in clinical trials of patients who have had relapses and who have been without treatment for at least 60 to 90 days may be another useful strategy. For example, the response to teniposide was higher in the presence of a drug-free interval longer than 2 months .
Better drugs are still needed, and new agents with novel mechanisms of action, as well as analogs of active drugs, should be tested. In this vein, the identification of paclitaxel and the topoisomerase I inhibitors as active is exciting, and incorporation of these into induction chemotherapy is a current research priority.
Overall and complete response rates seen with combination chemotherapy are generally superior to those observed with single agents in previously untreated patients, with the recent exception of oral etoposide in a prolonged schedule. The activity of the cytotoxic agents used, rather than the size of the dose, appears to be the greatest contributor to the therapeutic efficacy of any given combination regimen. Although an early randomized study  comparing a low-dose regimen of cyclophosphamide (500 mg/m²), methotrexate (10 mg/m²), and lomustine(Drug information on lomustine) (CeeNu, 50 mg/m²) (CMC) with a higher-dose regimen (cyclophosphamide, 1,000 mg/m²; methotrexate, 15 mg/m²; and lomustine, 100 mg/m²) showed a dose-response relationship, this was probably because CMC is not a very effective regimen and doses in the low-dose arm were lower than generally accepted as “standard.” Further investigation of dose intensity with more effective regimens in the outpatient setting has generally showed no dose-response relationship. A randomized comparison of high-dose versus conventional-dose cyclophosphamide, doxorubicin, and vincristine (CAV) and another trial of high-dose vs standard-dose cisplatin and etoposide (PE) in ED SCLC failed to show any evidence of superior response or prolonged survival with higher doses.
More recent investigations have focused on shortening the interval between treatments as opposed to giving higher doses less frequently. Although data from one study with highly selected ED patients is encouraging , thus far, no randomized trials have been reported that demonstrated an effect on survival for this approach.
In a series of trials investigating dose intensity with regimens requiring autologous marrow rescue, Johnson et al showed that while complete response rates can be increased, this does not translate into a survival benefit for patients with ED . Although more encouraging results with dose-intense regimens as late intensification therapy have been obtained in patients with LD, a randomized trial of this approach failed to produce a survival benefit in the dose-intense arm . However, with the availability of multilineage marrow growth factors and peripheral blood stem-cell support, the issue of dose intensity in SCLC continues to be an area of active research.
Many combinations of agents have been used as induction therapy for SCLC with similar results. Prior to the identification of etoposide as an active agent, CAV was one of the most commonly used regimens. However, the cardiac and mucosal toxic effects of doxorubicin caused problems, especially when the drug was combined with thoracic irradiation. Substitution of etoposide for doxorubicin in the CAV combination (CEV) was proven more effective and less toxic in a large randomized trial , and since the early 1980s, etoposide-based regimens have gradually supplanted CAV as induction therapy.
The activity of PE was first identified in refractory/recurrent disease, where it produced a 40% to 50% objective response rate with minimal toxicity . Based on these favorable results, this combination has been tested widely as first-line therapy and found to be at least as effective as other regimens and very well tolerated. Since the full-dose PE regimen lacks significant mucosal toxicity and is effective at minimally myelosuppressive dose levels, it can be safely administered with thoracic radiotherapy. This has increased median and long-term survival for patients with LD SCLC.
Despite the high rate of response to first-line chemotherapy, most patients with SCLC relapse with drug-resistant disease within months. Several approaches have been explored in an attempt to prevent this outcome. Based on treatment strategies for Hodgkin's disease and malignant lymphoma, it was thought that alternating two equally effective non-cross-resistant combinations early in treatment would lead to maximal antitumor effect (Goldie-Coldman principle). Based on the activity of PE in patients with disease refractory to or recurrent after CAV, these two regimens were tested in alternating and sequential strategies in several randomized studies [72-74]. Although one of these trials demonstrated a survival benefit from alternating therapy in a relatively small number of patients with LD , no improvement in outcome was identified for patients with ED. These results are probably due, in part, to the lack of bidirectional non-cross-resistance between CAV and PE.
Based on promising results obtained in a phase II pilot study of ifosfamide added to the base regimen of cisplatin and etoposide (VIP) in patients with ED, a phase III trial of the three-drug regimen vs PE was performed by the Hoosier Oncology Group . Long-term survival in the VIP arm was superior (13% vs 5% at 2 years), although overall and complete remission rates were comparable. In addition, the similar remission durations in both arms raises the issue of an imbalance of unknown prognostic factors contributing to the survival difference.
Most studies addressing the optimal duration of therapy have confirmed the efficacy of relatively short, intensive courses of treatment (4 to 6 months). Data from randomized trials indicate a somewhat longer duration of remission with continued chemotherapy, but without a detriment in survival duration for unmaintained patients if chemotherapy is given at relapse [76-78].
The activity of PE was first recognized in its use as a salvage regimen in patients with refractory or recurrent disease following initial doxorubicin-based therapy . After extensive evaluation of PE as induction therapy, however, it is clear that the partial non-cross-resistance is not bidirectional, with response rates to CAV of only 10% to 15% in patients failing to respond to or having recurrences after PE [72,74]. It is important to recognize, however, that the likelihood of response to any additional therapy is directly related to the drug-free interval. Patients who have relapses 2 years or more after induction therapy have higher response rates and longer median survivals than do patients who have relapses earlier .
Etoposide given orally over a protracted course produced response rates of 23% and 45% in two studies of relapsed SCLC patients, including those previously treated with intravenous etoposide [80,81]. However, responses were rare in the absence of a 3-month drug-free interval. This same schedule of etoposide as single-agent induction therapy produced response and survival rates similar to those of more toxic combination regimens in patients with ED and represents a reasonable treatment option for elderly or unfit patients [82-84].
After chemotherapy-induced complete remission, the chest remains a significant site of failure in patients with LD. In this context, several randomized trials have documented that the addition of radiotherapy to combination chemotherapy appears to reduce the incidence of relapses in the chest by 10% to 20% and to increase the complete response rate by 10% to 40% and the long-term survival rate by 5% to 15% . Much progress has been made recently in delineating the role of thoracic irradiation in the treatment of LD SCLC. Indeed, recognition of the benefit of thoracic radiotherapy, when properly combined with chemotherapy, can be viewed as one of the major advances in the treatment of this disease in the last decade. It must be recognized that concurrent chemoradiation is given at the expense of higher toxicity that can be prohibitive, especially when doxorubicin-containing regimens are used concomitantly with radiation. To minimize this problem, the minimally myelosuppressive and mucosa-sparing combination of PE has proven very valuable.
The timing of thoracic irradiation in the combined-modality treatment of LD SCLC is crucial. In general, most studies indicate that early administration of radiotherapy is superior to late treatment. In addition, a benefit from radiation has been difficult to demonstrate when it was given following completion of chemotherapy. Radiotherapy's early addition in an alternating or concurrent schedule appears to yield the best results, but it is also critical to maintain the dose intensity of both modalities [86,87].
Based on promising data using accelerated hyperfractionated radiotherapy in a pilot study, a large, randomized Intergroup trial testing its value vs conventional fractionation (both given concurrently with PE in the first cycle) was conducted. The preliminary results of this trial indicate no significant survival benefit in the hyperfractionated arm, although the duration of remission was approximately 2 months longer . Additionally, the accelerated therapy was associated with a higher incidence of acute esophageal toxic effects.
Nevertheless, the 2-year survival rate of more than 40% in both arms in this large Intergroup trial represents a significant improvement over previous results from large randomized trials in the United States. The favorable long-term outcome and the ability to deliver this regimen safely support the use of four cycles of PE with concurrent (conventional fractionation) thoracic radiotherapy in the first cycle as the standard of care for LD patients with disease amenable to a definitive combined approach.
Radiation to the chest as well as to sites of distant spread, such as the brain and bone, plays an important role in palliation of the patient with ED. Not unexpectedly, radiotherapy is not associated with a survival benefit in this setting.
Elective Brain Irradiation
Central nervous system metastases are an important cause of morbidity and mortality in patients with SCLC. These metastases have been reported in as many as 30% of all patients through the course of the disease and, in some series, in up to 80% of patients surviving 2 years or longer. Central nervous system metastases have occurred despite the use of aggressive combination chemotherapy. This has given rise to the belief that brain micrometastases are protected in a pharmacologic sanctuary. Hence, the concept of using elective brain irradiation to treat occult micrometastatic disease was developed.
Several randomized trials have demonstrated a significant reduction in SCLC relapse rates in the brain, from an average of 22% without elective brain irradiation to an average of 8% (a threefold reduction). Although none of the randomized trials have demonstrated an improvement in overall survival, there have been data that suggest a survival benefit for the subset of patients with a complete response to chemotherapy. The results of a randomized trial by Arriagada et al addressing this issue were recently published . Patients with LD or ED who had achieved a complete remission were randomly assigned to receive elective brain irradiation (24 Gy in 8 fractions) after chemotherapy) or no brain irradiation. The data from this trial of 294 patients reproduce results of earlier trials, demonstrating again a threefold reduction in the incidence of brain metastases in the irradiated group (40% vs 67% in the nonirradiated group at 2 years). Although disease-free survival was improved in the irradiated group, overall survival at 2 years did not differ significantly between the two arms. Similarly, a retrospective review of completely responding patients in one cooperative group's database yielded the same conclusion: a clear reduction in the rate of brain metastasis but no survival benefit with elective brain irradiation .
This absence of impact on survival has been a powerful argument against the use of elective brain irradiation, which is associated with late neurotoxicity, eg, organic brain syndromes, aphonia, hemiparesis, psychomotor retardation, and cerebellar dysfunction. The randomized trial by Arriagada et al is reassuring in that no differences were identified between the groups in neuropsychological function or abnormalities on CT scans, although follow-up is short. It also should be noted that the total dose used in that study was relatively low and that the chemotherapy and radiation were given at different times, two clearly significant variables influencing the risk of late neurotoxicity.
Although there are still no data demonstrating a survival impact for elective brain irradiation and it cannot be recommended routinely, it probably can be done safely, as demonstrated by the Arriagada et al trial. As therapy for SCLC evolves and more long-term survivors face the risk of late isolated brain SCLC recurrence, the question of the efficacy of elective brain irradiation will need to be readdressed.
Because of early metastasis to regional nodes and hematogenous dissemination, fewer than 10% of patients with SCLC are eligible for potentially curative surgical resection at the time of presentation. Even among this select subgroup of SCLC patients, surgery alone seldom results in long-term survival, with most major surgical series reporting a 5-year survival rate of 10% or less. However, if surgery is performed before chemotherapy or as an adjuvant treatment, it results in improved local disease control . The patients who appear to benefit most from a surgical approach are those without nodal involvement [92-94].
In one series , the median survival for patients with T1 or T2N0 tumors was 191 weeks using surgery and CAV chemotherapy, with a projected 5-year survival rate of 48%. On the other hand, the median survival is significantly shorter when nodes are involved, especially in the mediastinum. Because of this, thorough workup, including mediastinoscopy, is indicated for all patients with SCLC before surgical resection is considered.
Chemoprevention of Second Primary Tumors
Similar to patients who survive after treatment of squamous-cell cancer of the head and neck or non-small-cell lung cancer, patients with SCLC who survive longer than 2 years have a very high risk of developing second smoking-related primary tumors. Two recent retrospective analyses identified a 2% to 14% risk per year, with an actuarial cumulative risk of 70% 15 years from the start of treatment [95,96]. Most commonly, second primary tumors occur in the lung, with squamous-cell carcinoma the predominant histology. Unfortunately, very few patients reported in these two series were able to undergo successful surgical resection owing to the extent of tumor and medical contraindications. With the improving survival rates of patients with LD SCLC, second primary tumors are likely to become an increasingly common problem. Chemoprevention is an attractive option given the frequent inability to offer curative resection for the second cancer.
Previous data obtained by Hong et al  in patients with squamous-cell cancer of the head and neck and by Pastorino et al  in patients treated surgically for non-small-cell lung cancer indicate that retinoids may have activity in reducing the incidence of second primary cancers. In parallel with ongoing trials in patients treated for early stage head and neck cancers and non-small-cell lung cancers, a National Cancer Institute-sponsored phase III trial of 13-cis-retinoic acid vs placebo is in the planning stages for 3-year survivors of SCLC.