Combined Modality Therapy of Lung Cancer’

Combined modality therapy for lung cancer was first demonstrated to be successful in limited-stage small cell lung cancer. Concurrent administration of chemotherapy with chest and elective brain irradiation appears to produce the best results, with cisplatin/etoposide as the core chemotherapy. Using such programs, 2-year survival in the 40% range and 5-year survivals in excess of 20% may be expected, based on the results of multiple studies. Attempts to improve on these results through the use of altered schemes of chest irradiation or the delivery of high-dose consolidation chemotherapy are ongoing but to date have not been shown to affect survival significantly. We remain at a plateau in the effectiveness of combined modality therapy for small cell lung cancer, with little evidence that it impacts survival at all in extensive-stage disease. The incorporation of new agents in combination chemotherapy regimens, more “specific” immunotherapy directed at tumor-associated antigens, and the potential adjunctive use of broad-spectrum neuropeptide antagonists offer promise for the future. In non-small cell lung cancer, the sequential use of platinum-based chemotherapy and chest irradiation appears superior in survival to standard, daily fractionated radiation therapy used alone, with long-term survival increased from 5-10% to 15-20%. Concurrent administration of chemotherapy with cisplatin/etoposide and chest irradiation produces 2-year survival in the range of 30%, about twice that would be expected for radiation therapy alone, but has not been compared to it in the setting of a randomized trial. Low-dose cisplatin on a daily basis has been combined as a “sensitizer” with chest irradiation, producing initial results that appeared encouraging. However, these have not been reproduced in subsequent, randomized trials. Another approach to combined modalities has been to give chemotherapy or chemotherapy/radiation therapy as induction, followed by surgical resection, with or without subsequent additional treatment. Most patients (80-85%) can be resected, with encouraging survival at 2 and 3 years in the Southwest Oncology Group experience (37 and 26%, respectively). However, toxicity is greater, and such an apI Presented at “Foundations of Clinical Cancer Research: Perspective for the 21st Century.” a symposium to honor Emil J Freireich, M.D., on the occasion of his 70th birthday. March 14-15. 1997, M. D. Anderson Cancer Center. Houston, IX. 2 To whom requests for reprints should be addressed, at Division of Oncology. University of Washington. Box 356043. 1959 NE Pacific Street, Seattle. WA 98195-6043. Phone: (206) 548-4125: Fax: (206) 548-4509. proach is associated with an overall mortality risk in the range of 10%. A current intergroup study attempts to define the role of surgery in this setting. The major recent development that is likely to influence the future of combined modality therapy for this disease is the advent of multiple new chemotherapeutic agents, such as the taxanes, gemcitabine, vinorelbine, and the topoisomera-se-I inhibitors, which have activity in stage IV disease. The immediate challenge is how to combine these agents with platinum analogues, radiation, and surgery. Aiding this process may be the use of molecular biological “markers” that may predict the chance of success or failure with a given systemic agent. The next decade is likely to see substantial improvements in the outcome of treatment for patients with stages I-Ill non-small cell lung cancer, based on the systemic exploration of combined modalities. SCLC3 Sequential Approaches. When investigators began to use chemotherapy and RT together in the 1970s (1, 2), they chose the “sandwich” approach of chemotherapy radiotherapy chemotherapy, which had been pioneered with success in the treatment of pediatric cancers. Like these tumors, SCLC appeared to be “chemosensitive” as well as “radiosensitive.” The initial chemotherapy regimens involved agents that enhanced acute effects of RT on normal tissue via their systemic toxicities. Normal tissue damage could then be minimized by giving the modalities in sequence, rather than concurrently. This sequential combined approach appeared superior to RT used alone for patients with limited-stage disease but offered no apparent advantage over chemotherapy alone in extensive stage. In such programs for limited SCLC, the primary tumor was a common site of initial treatment failure. The reasons suggested were: inadequate field size; inadequate radiotherapy dose; and inherent tumor resistance. The SWOG addressed the issue of field size in a prospective, controlled trial (3). In that trial, patients with initial, partial response to combination chemotherapy were randomized to treatment with a large, “preinduction” field, which encompassed all of the tumor evident on X-ray before chemotherapy, or to the smaller “post-induction” treatment volume evident after chemotherapy. Although local failure was reduced by the use of the larger volume, there was no survival advantage. In fact, there were more 2-year disease-free 3 The abbreviations used are: SCLC, small cell lung cancer: NSCLC, non-small cell lung cancer; RI, radiotherapy; SWOG, Southwest Oncology Group; CR, complete response: CALGB. Cancer and Leukemia Group B: NCI, National Cancer Institute: CAV, cyclophosphamide/ doxorubicin/vincristine; PE, cisplatin/etoposide: PCI, prophylactic cranial irradiation: CNS, central nervous system; GM-CSF, granulocytemacrophage colony-stimulating factor: CBP, cyclophospharnide/ carmustine/cisplatin: PV, cisplatinlvinblastine; 5-FU, 5-fluorouracil; LCSG, Lung Cancer Study Group: RTOG, Radiation Therapy Oncology Group: RPSL, Rush-Presbyterian-St. Luke’s. Research. on April 15, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Clinical Cancer Research 2639 survivors in the reduced-field group (17% versus 7%). Although a randomized trial of RI dose has not been conducted, a SWOG trial with an investigator’s option of 3000 cGy versus 4500 cGy total dose demonstrated better 2-year survival for the higher dose (4). The same conclusion was reached by Cox et a!. (5) in a retrospective analysis. Using the sequential approach, four large randomized trials of chemotherapy alone versus combined modality treatment were reported in the 1980s. Three of these trials (3, 6, 7) found no advantage to RT plus chemotherapy over chemotherapy alone. Kies et a!. (3) in the SWOG trial focused on patients who had achieved radiographic CR after induction chemotherapy; the failure to demonstrate an advantage in this favorable subgroup led our group to abandon the “sandwich” model. The fourth study, reported by Perry et a!. (8) for the CALGB (8), demonstrated an advantage at 2 years for the sequential, combined modality approach versus chemotherapy alone (25% versus 15% at 2 years), but in that trial the “sequential” irradiation was delivered concurrently with the fourth cycle of chemotherapy. In addition, the local failure rate was still high in the combined modality arm (40%). Concurrent Approaches. The NCI-Navy group and its predecessor at the NCI-VA pioneered the use of concurrent, combined modalities in limited SCLC. In their randomized trial of doxorubicin-based chemotherapy versus the same chemotherapy plus initial, concurrent RT (9), improved median survival was seen with the combined approach (15 months versus 1 1.6 months), and a reduced incidence of failure at the primary site was observed (13% versus 29% chest alone recurrence). There was, however, a clinically significant increase in myelosuppression, esophagitis, pulmonary dysfunction, and weight loss with the addition of RI. In a trial of the Southeastern Cancer Study Group, Johnson et a!. ( 10) randomized patients to CAV versus CAV and RT, using a split-course regimen of RT (10). Median and 2-year survival rates favored the irradiated patients but were not statistically significantly different (14 months versus 13 months, 33% versus 24%, respectively). Again, the incidence of serious toxicities was higher in the combined modality group. A third randomized trial of initial concurrent chemotherapy with RI was that previously cited by CALGB (8), in which this approach was better than chemotherapy with cyclophosphamide, etoposide, and vincristine alone but slightly inferior to the concurrent, delayed approach (20% versus 25% at 2 years) for chemotherapy and RI. It should be noted that subsequent cycles of chemotherapy were markedly attenuated in the arm receiving initial chest RI, which may have impacted on the outcome. The key to improved results with the concurrent model lay in the development of a chemotherapy program with minimal normal tissue interaction with RI. This involved the combination of PE. Cisplatin is an agent that produces no stomatitis or esophagitis at normal doses and little myelosuppression. Its dose-limiting toxicities of nephropathy and vomiting can be managed effectively by modern techniques of hydration and antiemetic therapy, with sensory neuropathy as a remaining serious side effect but one which is related to cumulative dose and can be largely avoided if the agent’s use is confined to the induction period. Used alone, cisplatin had modest activity in SCLC (1 1), but the PE combination was reported to produce CR rates of 50% or greater in limited disease (12), as well as having some non-cross-resistance with CAV and other “standard” regimens (13). Etoposide is a very active single-agent in SCLC (14), and at “standard” doses has myelosuppression as its only common major toxicity. Cisplatin has weak radiosensitizing properties (15, 16), but much less so than doxorubicin, and neither component of PE has been clinically implicated in “radiation recall” toxicity. In a SWOG study, McCracken et a!. (17) combined PE with vincristine and concurrent, initial RI to the chest at 180 cGy per fraction to a total dose of 4500 cGy (17). At the completion of combined modality induction, consolidation chemotherapy was administered with vincristine plus methotrexate/etoposide alternating with doxorubicin and cyclophosphamide. Among 154 patients in this group-wide Phase II trial, CR occurred in 56%, with a median survival of 18 months. The 5-year survival was 25%, in contrast to results in the 10% range in our previous experience with chemotherapy or chemotherapy radiotherapy. This was followed by a randomized trial through NCI-Canada, reported by Murray et a!. (18), in which two approaches to concurrent, combined chemo(radio)therapy were tested: CAVIPE with “early” concurrent RI (first PE course); or CAV/PE with “delayed” concurrent RI (last PE course). The approach with “early” RI was superior with 5-year survival of 20% versus 1 1% (P = 0.008). Importantly, in this trial (as in SWOG), the early administration of chest RI did not compromise chemotherapy delivery; the dose intensity for each chemotherapy agent was nearly identical in the two arms. Intrathoracic toxicity to normal organs was minimized by the use of concurrent PE and RI. In McCracken’s study, grade 3/4 esophagitis occurred in 4% and grade 3 pneumonitis in 1% versus respective incidences of 66 and 37% in the NCI-Navy trial with doxorubicin-containing therapy (9). Indeed, even with sequential use of CAV and RI, Livingston et a!. (1) had observed respective incidences of 8 and 9%. By the early l990s, it was clear that concurrent, combined modality treatment of limited SCLC is superior to chemotherapy alone; that the intrathoracic toxicity of such treatment is reduced by coadmiistration of PE and RI, rather than earlier programs with more “interactive” drugs like doxorubicin and methotrexate; and that therapeutic efficacy is probably improved by early, concomitaut administration of chest irradiation and chemotherapy. Improving Chest Radiotherapy. An alternative approach to chest radiotherapy in the concurrent model was proposed by lumsi et al. (19), based on the observation that some lines of SCLC, grown in cell culture, demonstrate markedly increased sensitivity to small individual doses of radiation (loss of the “shoulder” on the radiation cell-survival curve) compared to other epithelial cells grown in culture (20). Such a situation should favor the use of individual fractions less than 2 Gy, given at shorter intervals than 24 h, because the difference between tumor tissue with a smaller or absent shoulder and normal tissue with a greater ability to repair such damage should be amplified. In a pilot study from the University of Pennsylvania, 150 cGy twice daily to the chest were given in 30 fractions over 3 weeks to a total dose of 4500 cGy, combined with PE chemotherapy (19). To minimize esophagitis, oblique fields were used in weeks 2 and 3, using computerized tomography for treatment planning. Subsequently, alternating chemotherapy with CAV/PE for six cycles and brain RI were administered. The CR rate from this program was 87%, with 2-year disease-free survival in excess of 50%. This trial, and another Research. on April 15, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 2640 Combined Modality Therapy of Lung Cancer 4 R. B. Livingston and J. Crowley, SWOG data, unpublished. encouraging pilot with similar design reported by Johnson et a!. (21, 22) for the NCI-Navy group led to a prospective, randomized intergroup trial. In this study, such “hyperfractionated” chest RI was compared to standard, once-daily continuous fractionation, with initial concurrent PE induction chemotherapy. Preliminary analysis of this trial has been presented by Johnson et a!. (23). Local control was improved by twice-daily RI, but the incidence ofesophagitis greater than to equal to grade 3 was also significantly higher. No statistically significant difference in survival was seen, although more mature analysis may demonstrate an advantage for the twice-daily approach. What is clear from this very large study is that concurrent use of PE and chest radiotherapy will produce 2-year survivals in the 40% range quite reproducibly. An even newer approach to optimization of chest radiotherapy in lung cancer involves three-dimensional, “conformal” treatment planning, which is made possible by modern techniques involving computerized tomography and exact simulation of target volume within the chest (24). Such treatment planning allows the delivery of unprecedented doses in the range of 70 Gy to the volume of interest, with a rapid falloff in the isodose plot such that surrounding normal structures do not receive substantially greater injury. Even with conformal therapy, however, the volume of normal tissue irradiated will depend on tumor size, and the tightness of margins will vary somewhat due to patient motion, breathing, and variation in the daily treatment setup. This is complex and labor intensive, and it remains to be seen whether conformal treatment planning will provide a practical advantage; no results have yet been reported in the combined modality setting. Elective PCI. It was recognized early in the treatment of SCLC that the brain and CNS represent a common “pharmacological sanctuary,” with resultant relapse at that site frequently identified. Following again the pediatric model of combined modality therapy as applied to acute lymphocytic leukemia, it became common practice to use PCI in an attempt to reduce the incidence of this devastating complication. It seems clear that PCI will reduce the likelihood of clinical relapse in the CNS, but it has little or no effect on the median survival in a disease where failure of control at other sites is the rule, nor does it affect the incidence of CNS metastases observed at autopsy (25, 26). Among 300 patients with SCLC who achieved CR, the cumulative risk of developing brain metastases was 45% for those not randomized to receive PCI versus 19% for those who received it (P < lO_6), with an overall cumulative risk of “isolated” CNS relapse of 40% (27). The subgroup that may be particularly likely to benefit from PCI are those patients with limited disease who achieve a CR. All series reporting 5-year survivals of 20% or greater have made use of PCI, which suggests but does not prove this point. Follow-up on the NCI-Navy study reported recently by Johnson et a!. (22), in which twice-daily chest RI was combined with PE induction, followed by CAV consolidation, makes a sobering point; although survival in this trial was 43% at 2 years, it had fallen to 19% at 5 years, and isolated CNS metastases caused more than 30% of the cancer deaths. About one-half of those with CNS relapse had received PCI. It was also recognized early that the use of PCI may be associated with the development of serious, debilitating late neurotoxicity. The incidence of this complication varies greatly from one retrospective series to another (28-31). It does appear that the risk is a function of dose per fraction and is higher with the use of individual fractions in excess of 200 cGy. Our current practice in the SWOG is to use a total dose of 3000 cGy, delivered at 200 cGy per fraction, over 3 weeks. What has only been recognized relatively recently is that patients with SCLC have a higher incidence of definable abnormalities in CNS function at presentation than do other cancer patients (including those with lung cancer; Ref. 32). This may be related to the high incidence of paraneoplastic syndromes in this disease and to cross-reactivity of the neuroendocrine tumor antigenically with neural tissue. In the trial by Arriagada et a!. (27), there were no significant differences in neuropsychological function or abnormalities detected on CI scan between those randomized to PCI and those who did not receive it. Improving Chemotherapy-Beyond PE. In the SWOG, we attempted to reduce the hematological toxicity and morbidity of the PE/chest RI program by the concurrent use of a hematopoietic growth factor (GM-CSF). As reported by Bunn et al. (33), there was an unexpected and clinically significant increase in the frequency and duration of life-threatening thrombocytopenia in patients randomized to GM-CSF (P < 0.001). There were also significantly more toxic deaths (P < 0.01), more nonhematological toxicities, more days in hospital, a higher incidence of the need for i.v. antibiotics, and the need for more transfusions in the GM-CSF group. In retrospect, it is likely that the GM-CSF mobilized peripheral blood stem cells (34), and that megakaiyocytic progenitors so mobilized were destroyed in increased numbers by the concurrently administered chest and brain radiotherapy. This trial effectively resulted in cessation of attempts to use growth factor support with concurrent chemo(radio)therapy programs. In turn, this has discouraged attempts at further dose escalation of chemotherapy during combined modality induction, because “standard” PE/RT produces a 24% incidence of grade 4 neutropema, with 22% hospitalized for i.v. antibiotics (33). What has been attempted is high-dose consolidation chemotherapy after induction chemo(radio)therapy. In another SWOG trial, Goodman et a!. (35) “intensified” patients with high-dose cyclophosphamide (150 mg/kg). Of 58 patients entered on the trial, only 21 (36%) actually received intensification, 7 of whom died of treatment-related causes. There was no evidence of an overall survival benefit. More recently, Elias et a!. (36) from Dana-Farber Cancer Institute reported encouraging results from high-dose consolidation with the combination of CBP, combined with autologous bone marrow transplantation. Among 19 patients who underwent CBP consolidation in remission, 53% were 2-year disease-free survivors. When one analyzes the results of the recently completed SWOG trials with concurrent PE/RI for induction, however, and includes only those patients who would have been eligible for the trial reported by Elias (CR or near CR, on study at 4 months), their survival appears to be comparable (51% 2-year disease-free).4 An intergroup Phase II pilot is now in progress, testing further the value of CBP with stem cell support after chemo(radio)-therapy induction. In the past several years, a number of new drugs have been identified with single-agent activity in extensive SCLC, including the taxanes (37) (paclitaxel and docetaxel) and topoisomerResearch. on April 15, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Clinical Cancer Research 2641 5 SWOG. unpublished data. ase1 interactive drugs related to camptothecin (Refs. 38 and 39; CPT1 1 and topotecan). None of these has yet been successfully incorporated into an induction program for limited SCLC. Other Approaches. Although the initial trials with “nonspecific” immunotherapy using agents like BCG and thymosin were disappointing (40, 41), hope remains that it may be possible to exploit a mechanism of specific antitumor immunity. It is known that lines derived from SCLC are deficient in the expression of class I MHC-restricted antigens (42), and that IFNs can up-regulate their expression in model systems (43). Because class I antigen expression is necessary to I cell-specific cytotoxicity, it was hoped that the administration of IFN might augment the host response to SCLC. Clinical trials of a-IFN have been either equivocal or negative to date (44, 45). Recently, Grant et a!. (46) reported preliminary results of another approach to the modification of biological response in which initial responders to combined modality induction were given “anti-idiotypic” therapy with BCG as an adjuvant plus a monoclonal antibody that mimics the GD3 ganglioside, a specific component of cells of neuroectodermal origin; four of six were alive at 3 years (three of six had extensive disease). A larger trial is necessary to confirm these findings. A new approach involves “designer” molecules that can antagonize SCLC-related autocrine growth factors. A multiplicity of such factors, especially neuropeptides, have been identified in SCLC (47, 48). A group of “broad spectrum” neuropeptide antagonists, which are analogues of substance P. have been identified and are able to inhibit SCLC growth in preclinical systems induced by a wide range of neuropeptides (49). One such compound blocks the growth of a SCLC xenograft in nude mice (50). Summary. Combined modality therapy with concurrent cisplatin-etoposide and chest radiotherapy has become the standard approach to the management of limited SCLC, with 2-year survival in the 40-45% range and 5-year survival in the 20-25% range. It is not yet clear whether twice daily RI will result in improved survival. PCI is widely applied today in the management of CR patients to prevent brain and CNS relapse and can reduce the incidence of this complication. It remains to be shown definitively that PCI improves survival, but this is suggested by its incorporation in the most successful programs. If it is to be used, PCI should be administered in low doses ( 200 cGy) per fraction, preferably separated in time from chemotherapy administration. We are presently at a plateau in effectiveness of the combined modality approach to this disease, with little evidence that it impacts survival at all in extensive stage patients, who unfortunately constitute the majority. Promising approaches to move us off the plateau include conformal chest radiotherapy, high-dose consolidation chemotherapy: the incorporation of new agents in combination chemotherapy regimens; more “specific” immunotherapy; and the adjunctive use of broad-spectrum neuropeptide antagonists. NSCLC-Combined Modalities In 1990, Dillman et a!. (5 1 ) for the CALGB reported a randomized trial of PV for two cycles followed by standard, continuous fractionated RI versus the same RI program without chemotherapy. Eligible patients had stage III disease by clinical or surgical staging, and patients were excluded with a history of weight loss more than 5%, positive supraclavicular nodes, or malignant pleural effusion. Induction chemotherapy with PV produced a 36% response rate, but overall response rates were not significantly different (56% versus 43%). However, survival was significantly improved in the combined arm (median of 14 months versus 10 months, P = 0.007) (Table 1). Dillman ci a!. (52) updated results of this study in 1996; the survival differences remained significant (24% versus 10% 3year, 17% versus 6% 5-year survival). This study was closed earlier than initially planned, with a total of only 155 patients, and 14% of those entered were excluded from analysis as ineligible. These and other concerns prompted a second randomized trial, reported by Sause et a!. (53). In that study, standard RI was compared to PV X 2 RI (“CALGB arm”) and to a third arm of hyperfractionated RI used alone (69.6 Gy at 1.2 Gy twice a day). Median survival was 11, 14, and 12 months for the three arms respectively, and 2-year survival was 19, 32, and 24% (P = 0.03). Staging in this trial was clinical, and patients with weight loss greater than 5% were again cxcluded. The power of this trial was greater ( 150 patients per arm). These two randomized studies led to the conclusion that combined modality therapy was superior to RI alone for the management of “good risk” patients with localized, but clinically inoperable NSCLC, and the use of PV X 2 RI remains an accepted community standard in such patients. Other controlled trials of the sequential model have been less positive. Le Chevalier et a!. (54) compared RI alone to the combination of vindesine, cyclophosphamide, cisplatin, and lomustine given for three cycles, followed by RI and three more cycles of chemotherapy. Staging was clinical, entry criteria were less rigid, and the difference in 2-year survival was of marginal significance (21% versus 14%, P 0.08), although the mcidence of distant metastasis was markedly reduced in the cornbined group. The response rate to initial chemotherapy was 27%. The SWOG compared RI alone to an alternating chemotherapy program followed by RI [5-FU, vincristine, mitomycin/cyclophosphamide, doxorubicin, and cisplatin], and found no evidence of a survival advantage for the combined modalities.5 Staging was clinical, and entry criteria were generous. Whether the positive results for the CALGB regimen are due to greater activity of PV than the other regimens or more strict selection criteria remain an open question, but these observations opened the door to vigorous exploration of combined modalities over the past decade. A second sequential approach involves chemotherapy followed by surgery, rather than RI, as the local modality. Martini et a!. (55) summarized the experience at Memorial Hospital with a regimen of mitomycin C, vindesine or vinblastine, and cisplatin given for two or three cycles to patients with clinical IIIaN, disease (ipsilateral involvement of mediastinal nodes). Among 136 treated, the response rate to induction chemotherapy was 77%; the resectability rate was 65%, with pathological CR in 14% of all patients (21% of those resected). Threeand 5-year survivals were 28 and 17%, respectively, with treatment-related deaths in 5%. Burkes et a!. (56) reported a trial from Toronto in Research. on April 15, 2017. © 1997 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 2642 Combined Modality Therapy of Lung Cancer Tab!e 1 Sequential model: Che motherapy foll owed by local treatment in stage III NSCLC Localarm Response (%) MST” (mo) Su rvival (%) at