Tumor Cell Lines Established in Vitro: An Independent Prognostic Factor for Survival in Non-Small-Cell Lung Cancer Henry Stevenson, MD; Adi F. Gazdar, MD; Ruby Phelps, AAS; R. Ilona Linnoila, MD; Daniel C. Ihde, MD; Bimal Ghosh, MD; Thomas Walsh, MD; Edward L. Woods, MD; Herbert Oie, PhD; Teresa O'Connor, MPH; Robert Makuch, PhD; Barnett S. Kramer, MD; and James L. Mulshine, MD
Objective: To determine the relation between in-vitro establishment of tumor cell lines and survival in patients with non-small-cell lung cancer. Design: Cohort study. Setting: Single-institution tertiary care center. Patients: One hundred twenty-three consecutive patients with non-small-cell lung cancer from whom a viable tumor specimen could be obtained. Intervention: Tumor tissue was removed at the time of entry into a therapeutic protocol. The tumor tissue was processed in the laboratory for attempted cell-line establishment. Patients classified as potentially curable (stages I, II, and HIA) were treated with surgical resection, radiation therapy, or a combination. Patients suitable for palliative therapy only (stages HIB and IV) were treated with radiation therapy with or without chemotherapy. Chemotherapy was based on in-vitro drug sensitivity when available. Cell-line establishment was correlated to clinical outcome. Measurements and Main Results: Univariate analysis of survival was done using the log-rank test; multivariate analysis was done by Cox modeling step-up and step-down techniques. Cell lines were established from the tumor specimens of 25 patients (20%). Those patients experienced a median survival of 7 months compared with 18 months in patients from whom cell lines could not be established (P < 0.001). In the 61 patients with potentially curable disease, 8 patients (13%) with cell lines established had a median survival of 8 months compared with 32 months for those without cell lines established (P = 0.001). In the 62 palliative group patients, the median survival of the 17 patients (27%) from whom tumor cell lines were established was 5 months compared with 7 months for those without cell lines (P = 0.15). Multivariate analysis in both groups showed cell-line establishment to be a significant indicator of prognosis (P < 0.0001 for curable group; P < 0.01 for palliative group). Conclusion: In-vitro tumor growth is related to decreased patient survival, which in turn reflects the biologic aggressiveness of cancers giving rise to these tumor cell lines.
Annals of Internal Medicine. 1990;113:764-770. From the National Cancer Institute, National Naval Medical Center, and Uniformed Services University of the Health Sciences, Bethesda, Maryland; and Yale University, New Haven, Connecticut. For current author addresses, see end of text. 764
Approximately 120 000 new cases of non-small-cell lung cancer (squamous, large-cell, and adenocarcinoma) will be diagnosed in the United States in 1990, and only 15% of these patients can be cured with the best treatments (1). Approximately 70% of patients have spread of disease to mediastinal nodes or distant sites by the time of initial diagnosis, with fewer than 5% exhibiting long-term, disease-free survival. The potential for cure is highest in patients with localized, resectable disease confined to the pulmonary parenchyma and hilar lymph nodes (1, 2). In a substantial percentage of these earlystage patients, however, the tumor metastasizes and the patients die after definitive local treatment (3, 4). Such patients with occult metastatic disease are theoretically the most favorable candidates for adjuvant chemotherapy (5). For this reason, investigators have commented on the need for better ways of identifying patients who may benefit from adjuvant interventions (6). We have had a long-standing interest in the propagation of tumor-containing, small-cell lung cancer specimens and are currently able to initiate cell lines in 40% of clinical cases (7). This ability has permitted extensive studies of the biologic behavior of small-cell lung cancer (8-10). As a pilot effort, we designed a clinical protocol for patients with non-small-cell lung cancer from whom viable tumor tissue could be obtained for in-vitro cell culture and attempted to establish a continuous cell line. This trial provided an opportunity to evaluate the feasibility and utility of prospective in-vitro chemotherapy sensitivity analysis (11, 12). Our results, based on our experience with the first 123 patients in this study, indicate that successful cell-line generation is an important negative prognostic factor for survival, independent of other known prognostic factors, in patients with nonsmall-cell lung cancer.
Patients and Methods Staging and Treatment One hundred and twenty-seven patients with any stage or histologic specimen of pathologically confirmed non-small-cell lung cancer with an Eastern Cooperative Oncology Group (ECOG) performance status of 3 or better were entered into our study between May 1984 and January 1988. The schema for this study is shown in Figure 1. This prospective clinical trial was approved by the institutional review board. Viable tumor tissue from all but 4 patients was processed in our laboratory as part of the protocol entry process. Specimens from two patients did not contain discernible viable tumor, and the specimens from two other patients were inad-
15 November 1990 • Annals of Internal Medicine • Volume 113 • Number 10
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19716/ by a University of California San Diego User on 02/03/2017
Figure 1. Methods flow diagram outlining the tumor tissue specimens and management techniques of patients with nonsmall-cell lung cancer used in our study. vertently submitted in formalin; these four patients were excluded from all subsequent analyses. Tumor specimens from patients having procedures requiring general anesthesia (such as mediastinoscopy or thoracic resection) were procured only during standard staging or therapeutically indicated procedures. Biopsies for the exclusive purpose of tumor tissue acquisition for protocol entry (including peritoneoscopic-directed biopsy of liver lesions, biopsy of peripheral lymph nodes, and aspirations of pleural effusions) were allowed if they could be done under local anesthesia. Liver biopsies were done if imaging studies showed parenchymal lesions in patients without other sites of accessible tumor involvement. Drug sensitivity testing was done on fresh tumor tissue or cell lines grown from fresh tumor specimens. Chemotherapy, which was given only when clinically indicated, was selected on the basis of in-vitro test results whenever possible. Preliminary clinical correlations of in-vitro biologic features, drug sensitivity, and clinical outcome were reported previously (12). We obtained chest roentgenograms and chest computed tomography (CT) scans, with or without contrast, on all patients. Radionuclide scans of liver or bone were done on patients with abnormal liver function test results, hepatomegaly, elevated alkaline phosphatase levels, or bone pain. Head CT scans were done on patients with neurologic signs and symptoms. Patients with otherwise localized tumor and clinically suspicious lymph nodes on chest roentgenograms or chest CT scans were routinely evaluated surgically for nodal involvement of suspicious sites, before resection of the primary lesion. The presence or absence of clinically detectable metastatic disease was specified for each patient. When this study was started in 1984, patients were staged according to the then-contemporary American Joint Committee (AJC) guidelines (13) and placed, for purposes of analysis, in one of two treatment groups: potentially curable or suitable only for palliative therapy. For the current analysis, the initial staging classification was modified to conform to the new AJC staging system (14). Potentially curable patients (stage I, II, and III A patients, excluding what is now designated stage IIIB) had tumor confined to the thorax (14). Patients in this group had either surgical resection (with or without postoperative involved field radiotherapy) or definitive radiotherapy using standard indications (11). On relapse, any patient with a good performance status and disease not requiring palliative radiotherapy was oflfered chemotherapy according to the best combination suggested by previous drug sensitivity assay results. If drug sensitivity information was not available, the
combination of etoposide and cisplatin was offered because of the favorable experience of large cooperative groups with this outpatient regimen (15, 16). The palliative treatment group included patients presenting with regionally advanced intrathoracic disease or supraclavicular nodes (stage IIIB) and patients with distant metastases (stage IV). Patients with good performance status and measurable or evaluable disease not requiring palliative irradiation were oflfered immediate chemotherapy with the best combination suggested by the in-vitro sensitivity assay results, or etoposide and cisplatin if drug sensitivity results were not available. Patients with tumors that demonstrated evidence of neuroendocrine differentiation based on results from a battery of in-vitro tests (elevated DOPA-decarboxylase activity, elevated creatine kinase activity, or presence of neurosecretory granules on electron microscopy) were treated similarly (12), with the exception that patients in the palliative group were initially offered chemotherapy with cyclophosphamide, methotrexate, CCNU, vincristine, doxorubicin, and procarbazine (CMC and VAP), which was the standard combination chemotherapy used at our institution in 1984 for small-cell lung cancer. The treatment for this group was later changed to etoposide and cisplatin. The trial design was simplified because of the subsequent recognition of at least equivalent treatment outcome and reduced toxicity using etoposide and cisplatin instead of CMC and VAP in our patients with small-cell lung cancer (17). Patients received regular follow-up and were restaged at relapse. Standard chemotherapy response criteria, applied to
Table 1. Patient
Charexcteristi cs at Stu\dy En
Objective: To determine the relation between in-vitro establishment of tumor cell lines and survival in patients with non-small-cell lung cancer. Design: Cohort study. Setting: Single-institution tertiary care center. Patients: One hundred twenty-three consecutive patients with non-small-cell lung cancer from whom a viable tumor specimen could be obtained. Intervention: Tumor tissue was removed at the time of entry into a therapeutic protocol. The tumor tissue was processed in the laboratory for attempted cell-line establishment. Patients classified as potentially curable (stages I, II, and HIA) were treated with surgical resection, radiation therapy, or a combination. Patients suitable for palliative therapy only (stages HIB and IV) were treated with radiation therapy with or without chemotherapy. Chemotherapy was based on in-vitro drug sensitivity when available. Cell-line establishment was correlated to clinical outcome. Measurements and Main Results: Univariate analysis of survival was done using the log-rank test; multivariate analysis was done by Cox modeling step-up and step-down techniques. Cell lines were established from the tumor specimens of 25 patients (20%). Those patients experienced a median survival of 7 months compared with 18 months in patients from whom cell lines could not be established (P < 0.001). In the 61 patients with potentially curable disease, 8 patients (13%) with cell lines established had a median survival of 8 months compared with 32 months for those without cell lines established (P = 0.001). In the 62 palliative group patients, the median survival of the 17 patients (27%) from whom tumor cell lines were established was 5 months compared with 7 months for those without cell lines (P = 0.15). Multivariate analysis in both groups showed cell-line establishment to be a significant indicator of prognosis (P < 0.0001 for curable group; P < 0.01 for palliative group). Conclusion: In-vitro tumor growth is related to decreased patient survival, which in turn reflects the biologic aggressiveness of cancers giving rise to these tumor cell lines.
Annals of Internal Medicine. 1990;113:764-770. From the National Cancer Institute, National Naval Medical Center, and Uniformed Services University of the Health Sciences, Bethesda, Maryland; and Yale University, New Haven, Connecticut. For current author addresses, see end of text. 764
Approximately 120 000 new cases of non-small-cell lung cancer (squamous, large-cell, and adenocarcinoma) will be diagnosed in the United States in 1990, and only 15% of these patients can be cured with the best treatments (1). Approximately 70% of patients have spread of disease to mediastinal nodes or distant sites by the time of initial diagnosis, with fewer than 5% exhibiting long-term, disease-free survival. The potential for cure is highest in patients with localized, resectable disease confined to the pulmonary parenchyma and hilar lymph nodes (1, 2). In a substantial percentage of these earlystage patients, however, the tumor metastasizes and the patients die after definitive local treatment (3, 4). Such patients with occult metastatic disease are theoretically the most favorable candidates for adjuvant chemotherapy (5). For this reason, investigators have commented on the need for better ways of identifying patients who may benefit from adjuvant interventions (6). We have had a long-standing interest in the propagation of tumor-containing, small-cell lung cancer specimens and are currently able to initiate cell lines in 40% of clinical cases (7). This ability has permitted extensive studies of the biologic behavior of small-cell lung cancer (8-10). As a pilot effort, we designed a clinical protocol for patients with non-small-cell lung cancer from whom viable tumor tissue could be obtained for in-vitro cell culture and attempted to establish a continuous cell line. This trial provided an opportunity to evaluate the feasibility and utility of prospective in-vitro chemotherapy sensitivity analysis (11, 12). Our results, based on our experience with the first 123 patients in this study, indicate that successful cell-line generation is an important negative prognostic factor for survival, independent of other known prognostic factors, in patients with nonsmall-cell lung cancer.
Patients and Methods Staging and Treatment One hundred and twenty-seven patients with any stage or histologic specimen of pathologically confirmed non-small-cell lung cancer with an Eastern Cooperative Oncology Group (ECOG) performance status of 3 or better were entered into our study between May 1984 and January 1988. The schema for this study is shown in Figure 1. This prospective clinical trial was approved by the institutional review board. Viable tumor tissue from all but 4 patients was processed in our laboratory as part of the protocol entry process. Specimens from two patients did not contain discernible viable tumor, and the specimens from two other patients were inad-
15 November 1990 • Annals of Internal Medicine • Volume 113 • Number 10
Downloaded From: http://annals.org/pdfaccess.ashx?url=/data/journals/aim/19716/ by a University of California San Diego User on 02/03/2017
Figure 1. Methods flow diagram outlining the tumor tissue specimens and management techniques of patients with nonsmall-cell lung cancer used in our study. vertently submitted in formalin; these four patients were excluded from all subsequent analyses. Tumor specimens from patients having procedures requiring general anesthesia (such as mediastinoscopy or thoracic resection) were procured only during standard staging or therapeutically indicated procedures. Biopsies for the exclusive purpose of tumor tissue acquisition for protocol entry (including peritoneoscopic-directed biopsy of liver lesions, biopsy of peripheral lymph nodes, and aspirations of pleural effusions) were allowed if they could be done under local anesthesia. Liver biopsies were done if imaging studies showed parenchymal lesions in patients without other sites of accessible tumor involvement. Drug sensitivity testing was done on fresh tumor tissue or cell lines grown from fresh tumor specimens. Chemotherapy, which was given only when clinically indicated, was selected on the basis of in-vitro test results whenever possible. Preliminary clinical correlations of in-vitro biologic features, drug sensitivity, and clinical outcome were reported previously (12). We obtained chest roentgenograms and chest computed tomography (CT) scans, with or without contrast, on all patients. Radionuclide scans of liver or bone were done on patients with abnormal liver function test results, hepatomegaly, elevated alkaline phosphatase levels, or bone pain. Head CT scans were done on patients with neurologic signs and symptoms. Patients with otherwise localized tumor and clinically suspicious lymph nodes on chest roentgenograms or chest CT scans were routinely evaluated surgically for nodal involvement of suspicious sites, before resection of the primary lesion. The presence or absence of clinically detectable metastatic disease was specified for each patient. When this study was started in 1984, patients were staged according to the then-contemporary American Joint Committee (AJC) guidelines (13) and placed, for purposes of analysis, in one of two treatment groups: potentially curable or suitable only for palliative therapy. For the current analysis, the initial staging classification was modified to conform to the new AJC staging system (14). Potentially curable patients (stage I, II, and III A patients, excluding what is now designated stage IIIB) had tumor confined to the thorax (14). Patients in this group had either surgical resection (with or without postoperative involved field radiotherapy) or definitive radiotherapy using standard indications (11). On relapse, any patient with a good performance status and disease not requiring palliative radiotherapy was oflfered chemotherapy according to the best combination suggested by previous drug sensitivity assay results. If drug sensitivity information was not available, the
combination of etoposide and cisplatin was offered because of the favorable experience of large cooperative groups with this outpatient regimen (15, 16). The palliative treatment group included patients presenting with regionally advanced intrathoracic disease or supraclavicular nodes (stage IIIB) and patients with distant metastases (stage IV). Patients with good performance status and measurable or evaluable disease not requiring palliative irradiation were oflfered immediate chemotherapy with the best combination suggested by the in-vitro sensitivity assay results, or etoposide and cisplatin if drug sensitivity results were not available. Patients with tumors that demonstrated evidence of neuroendocrine differentiation based on results from a battery of in-vitro tests (elevated DOPA-decarboxylase activity, elevated creatine kinase activity, or presence of neurosecretory granules on electron microscopy) were treated similarly (12), with the exception that patients in the palliative group were initially offered chemotherapy with cyclophosphamide, methotrexate, CCNU, vincristine, doxorubicin, and procarbazine (CMC and VAP), which was the standard combination chemotherapy used at our institution in 1984 for small-cell lung cancer. The treatment for this group was later changed to etoposide and cisplatin. The trial design was simplified because of the subsequent recognition of at least equivalent treatment outcome and reduced toxicity using etoposide and cisplatin instead of CMC and VAP in our patients with small-cell lung cancer (17). Patients received regular follow-up and were restaged at relapse. Standard chemotherapy response criteria, applied to
Table 1. Patient
Charexcteristi cs at Stu\dy En
