Presented as a poster discussion (abstract 7527) at the 2008 Annual Meeting of the American Society of Clinical Oncology, Chicago, Illinois, May 30 to June 3, 2008.
Although chemotherapy and radiation therapy currently are recommended in limited-stage small cell lung cancer (L-SCLC), several small series have reported favorable survival outcomes in patients who underwent surgical resection. The authors of this report used a US population-based database to determine survival outcomes of patients who underwent surgery.
The Surveillance, Epidemiology, and End Results (SEER) registry was used to identify patients who were diagnosed with L-SCLC between 1988 and 2002 coded by SEER as localized disease (T1-T2Nx-N0) or regional disease (T3-T4Nx-N0). Kaplan-Meier and Cox regression analyses were used to compare overall survival (OS) for all patients.
In total, 14,179 patients were identified, including 863 patients who underwent surgical resection. Surgery was associated more commonly with T1/T2 disease (P < .001). Surgery was associated with improved survival for both localized disease and regional disease with improvements in median survival from 15 months to 42 months (P < .001) and from 12 months to 22 months (P < .001), respectively. Lobectomy was associated with the best outcome (P < .001). Patients with localized disease who underwent lobectomy with had a median survival of 65 months and a 5-year OS rate of 52.6%; whereas patients who had regional disease had a median survival of 25 months and a 5-year OS rate of 31.8%. On multivariate analysis, the benefit of surgery varied in a time-dependant fashion. However, the benefit of lobectomy remained across all time intervals (P = .002).
There are approximately 215,000 new cases of lung cancer each year,1 and approximately 13% of those are small cell lung cancers (SCLC).2 SCLC behaves differently from the more common nonsmall cell lung cancers (NSCLC) and is noteworthy for its rapid growth and early development of widespread metastases.3, 4 Despite initial responses to therapy, most patients with SCLC develop local recurrence.5
Historically, surgery was used to treat this disease. However, the poor survival reported in a Medical Research Council (MRC) trial6 performed in the 1960s led to the abandonment of surgery as a standard treatment in favor of chemotherapy. Two subsequent meta-analyses7, 8 revealed that the addition of thoracic radiation to systemic chemotherapy improved survival, and that has become the standard of care.
Recently, several institutions have reported positive experience with surgery in patients with early stage disease,9-17 and other series have reported favorable results with surgical intervention in stage III disease as well.15-18 Therefore, we analyzed data from the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute (Bethesda, Md) to determine the role of surgery in a large cohort of patients with SCLC from 1988 to 2003.
MATERIALS AND METHODS
The SEER database is a National Cancer Institute program that serves as the representative cancer registry of the United States. SEER is composed of a set of geographically defined, population-based central cancer registries in the United States that collect data concerning individual patient demographics, diagnosis, treatment, and survival outcomes. Since 2001, these data have been collected from 17 regions that cover 26% of the US population (available at: http://www.seer.cancer.gov/about; accessed April 29, 2008). Because the individual patient data are deidentified, approval from an ethics committee or institutional review board is not required.
We identified patients ages 20 years to 75 years who were diagnosed from 1988 through 2002 with nonmetastatic SCLC. The histologies that we included were SEER codes 8002/3 (malignant tumor, small cell type), 8041/3 (small cell carcinoma, not otherwise specified [NOS]), 8042/3 (oat cell carcinoma), 8043/3 (small cell carcinoma, fusiform cell), and 8044/3 (small cell carcinoma, intermediate cell). In patients for whom it was unknown whether or not they underwent surgery (n = 1883), those who had SEER codes 8246/3 (neuroendocrine carcinoma, NOS; n = 977) or 8045/3 (combined small cell lung cancer; n = 228) were excluded. In total, 14,179 patients were identified for the current analysis.
By using the SEER coding for extent of disease, patients were grouped into those with “localized” disease or “regional” disease (available at: http://www.seer.cancer.gov/manuals/EOD10Dig.third.pdf accessed on April 29, 2008). Localized disease was correlated with a TNM classification of T1-T2Nx-N0, and regional disease was correlated with a classification of either T3-T4Nx-N0 or T1-T4N1-N2. We excluded patients who were coded with contralateral lung disease, separate tumor nodules in a different lobe, or metastasis. Patients with N3 disease were coded by SEER as metastatic and also were excluded.
Patients who were included in this study were identified as having received or not received external beam radiation therapy. Patients who had adjuvant radiation therapy recommended but refused it, who received brachytherapy, or who had unknown status regarding the receipt of radiation therapy were not included in the analysis. Curative surgery was defined as pneumonectomy, lobectomy, or sublobar resection. Chemotherapy data are not available in the SEER database.
The primary endpoint of this study was all-cause mortality. Follow-up was calculated from the month and year of initial diagnosis and was available through December 2004 at the time of our analysis. Vital status and the date of last contact were available for all patients.
Kaplan-Meier analysis with the log-rank test was used to plot survival curves for comparisons between patients who underwent surgery and those who received either radiation therapy or no additional locoregional treatment and to determine the extent of differences among the 4 surgical groups without adjustment for covariates. Univariate and multivariate Cox regression analyses were used to examine possible confounding and interactive effects of surgical group with the following predictors: radiation therapy (yes or no), lymph node status (N0, N1, or N2), tumor classification (T1-T2, T3, or T4), age, race (black, white, or other), and sex. To determine the effect of age on survival, patients were divided into age deciles, and the relative hazard of death was measured within each decile. Because we observed that the hazard increased in a sufficiently monotonic and even manner, we considered it justified to use age as a continuous covariate; other predictors were considered categorical. Interactions of surgical group with the other predictors were tested, and the significant terms were retained in the final model. The supremum test of Lin et al19 was used to assess conformity with the proportional hazards assumption, and separate hazard ratio estimates were fitted for different time intervals when necessary. Comparisons of the characteristics between the surgery and no-surgery groups were made using Pearson chi-square tests. Statistical analyses were performed using SPSS software version 16.0 (SPSS Inc., Chicago Illinois) and SAS release 9.2 (SAS Institute, Cary, NC). Statistical significance was defined as a 2-sided P value ≤.05.
Among the 14,179 patients who were included in the current study, the median age was 65 years (interquartile range, 58-70 years), the median follow-up was 52 months, and 1678 patients were alive at time of analysis. Regarding treatment, 863 patients (6.1%) underwent surgery, and 241 of those patients also received postoperative radiation therapy (PORT). Of the patients who underwent surgery, 515 underwent lobectomy, 277 underwent sublobar resection, and 71 underwent pneumonectomy. Among the 515 patients who underwent lobectomy, 246 patients had localized disease, and 269 patients had regional disease. Table 1 provides a comparison of characteristics by treatment assignment. Patients who underwent surgery were more likely to have localized disease.
Table 1. Patient Characteristics Stratified by Use of Surgery
Data are presented as number of patients, with percentages in parentheses unless noted otherwise.
Comparisons between the surgery and no surgery groups were compared using the Pearson chi-square test.
Use of Surgery and Survival
For the whole cohort, surgery was associated with a median survival of 28 months compared with 13 months for no surgery and with a 5-year overall survival (OS) rate of 34.6% versus 9.9% (P < .001). In a subgroup analysis by disease stage, the median survival for patients with localized disease was 42 months versus 15 months in favor of surgery with a 5-year OS rate of 44.8% versus 13.7% (P < .001). For patients with regional disease, the median survival was 22 months versus 12 months in favor of surgery with a 5-year OS rate of 26.3% versus 9.3% (P < .001).
Effect of Surgery Type on Survival
Comparing lobectomy, pneumonectomy, sublobar resection, and no surgery, all 4 survival distributions differed from every other distribution (P < .001) even after a Tukey adjustment for multiple tests. The median time to death (with 95% confidence interval [CI]) was 13 months (95% CI, 12-13 months) for no surgery, 20 months (95% CI, 15-33 months) for pneumonectomy, 23 months (95% CI, 19-25 months) for sublobar resection, and 40 months (95% CI, 32-50 months) for lobectomy (Fig. 1). Patients who underwent lobectomy with localized disease had a median survival of 65 months and a 5-year OS rate of 52.6%. Patients with regional disease had a median survival of 25 months and a 5-year OS rate of 31.8%.
Effect of Surgery by Lymph Node Status
A subgroup analysis was performed to assess surgical outcomes based on lymph node (N) status. In total, 11,207 patients had enough information regarding lymph node disease and location of the positive lymph nodes to group them into the appropriate lymph node status according to the sixth edition of the American Joint Committee on Cancer (AJCC) Cancer Staging Manual.20 Two thousand two hundred fifty-one patients had N0 disease, 802 patients had N1 disease, and 7974 patients had N2 disease. For N0 disease, 435 patients underwent surgery, 1816 patients did not, and the median survival was 40 months versus 15 months in favor of the surgery group (P < .001) (Fig. 2). For N1 disease, 164 patients underwent surgery, 638 patients did not, and the median survival was 29 months versus 14 months in favor of the surgery group (P < .001) (Fig. 3). For N2 disease, 187 patients underwent surgery, 7787 patients did not, and the median survival was 19 months versus 12 months in favor of the surgery group (P < .001) (Fig. 4).
Use of Postoperative Radiation Therapy
Of 863 patients who underwent surgery, 241 patients received postoperative treatment. Most patients who received PORT had regional disease (72.2%). Comparing PORT with surgery alone across the whole cohort, the median survival was 31 months for surgery alone versus 26 months for PORT (P = .06). However, on subgroup analysis, a benefit from postoperative radiation emerged based on lymph node status. For patients with N0 and N1 disease, there was no benefit from PORT: The median survival was 42 months for surgery alone versus 41 months for PORT in patients with N0 disease (P = .44), and the median survival was 35 months for surgery alone versus 22 months for PORT in patients with N1 disease (P = .179). However, for patients with N2 disease, there was a statistically significant survival benefit from PORT: The median survival was 16 months for surgery alone versus 22 months for PORT (P = .011) (Fig. 5).
Increased age and being a man were associated significantly with a poor prognosis across all time intervals. We noted that other variables had various effects on survival across the time of follow-up. To generate more reliable hazard ratios, the period of follow-up was divided arbitrarily into 3 blocks: from 0 months to 30 months, from 31 months to 60 months, and >60 months. We observed that the effect of surgery interacted significantly with the effect of radiation therapy during Block 1 and with the effect of lymph node status during Blocks 1 and 2. For this reason, we present the effect of surgery stratified by radiation therapy use and lymph node status only for Blocks 1 and 2. The hazard ratios, CIs, and P values from the multivariate analysis are shown in Table 2. The impact of any surgery on survival generally was favorable up to 2.5 years. The protective effect of lobectomy appeared to be fairly consistent throughout the period of follow-up with hazard ratios from 0.32 to 0.84 relative to no surgery.
Table 2. Multivariate Analysis of Survival Comparing Surgery Versus No Surgery
HR indicates hazard ratio; CI, confidence interval; RT, radiotherapy; N, American Joint Committee on Cancer lymph node status.
Because lymph node status and receipt of RT did not interact with surgery after 5 years, the hazard ratios are presented without any stratification.
The results from this large, population-based cohort indicated that patients who underwent surgical intervention had a 5-year OS rate of 34.6%. This benefit was noted particularly in patients with localized disease, corresponding to AJCC stage I, with a 5-year OS rate of 44.8%. In patients with regional disease, the survival rate was worse at 26.3% but still compared favorably with historic controls. The patients who underwent lobectomy had the best outcomes with 5-year OS rates of 52.6% for those with localized disease and 31.8% for those with regional disease.
Currently, surgery generally is not recommended as standard management in patients with limited-stage SCLC. However, the National Comprehensive Cancer Network does recommend surgery followed by adjuvant chemotherapy as standard management for T1-T2N0 disease after a full staging workup.21 For all other stages (excluding M1), chemoradiation is recommended. Several influential trials led to the abandonment of surgery in the treatment of this disease. The MRC conducted a randomized trial of surgery versus radiation alone.6 In that trial, the median survival in the surgery arm was 6.5 months versus 10 months in the radiation arm (P = .04). However, criticism of that trial included the finding that only 34 of the 71 patients (48%) who were enrolled in the surgery arm actually underwent surgery, which may have diluted the survival results from patients in that group.
A second influential trial was published in 1994 by the Lung Cancer Study Group (LCSG).22 In that study, 146 patients with limited-stage SCLC who had regional lymph node involvement (T3N1-N2M0) received 5 cycles of neoadjuvant chemotherapy (combined cyclophosphamide, doxorubicin, and vincristine) and then were randomized to surgery followed by PORT versus definitive radiation alone. The study revealed that the 2-year OS rate was 20% in each arm and that there was no role for surgery in the multimodality treatment of SCLC. However, in that study, all patients in the surgical arm received PORT, but only 32% of them were staged pathologically with N2 or unresectable disease. Although there is scant literature on PORT for patients with SCLC, based on the NSCLC literature on PORT, there may be a detriment to survival in adding thoracic radiation to the treatment of patients with N0-N1 disease.23, 24
An additional critique of the LCSG study is that the trial was conducted before the widespread use of platinum-based chemotherapy, which occurred during the accrual of the study in the late 1980s. There are data suggesting that survival with surgery is improved with platinum-based chemotherapy. Brock et al16 recently reported on their institutional experience in surgically treating 82 patients with SCLC in the platinum era. Among their findings was that, when comparing the 56 patients who received neoadjuvant or adjuvant chemotherapy, the 5-year OS rate was superior in the platinum-containing arm (62% vs 36%; P = .05).
In the current series, we did not have access to chemotherapy data. However, the 5-year OS rates of 44.8% for patients with limited disease and 26.3% for patients with regional disease who underwent surgery are within the range reported in the other single-institution series in which patients were treated with chemotherapy. Badzio et al performed a matched-pair analysis25 comparing 67 patients who underwent surgery followed by adjuvant chemotherapy versus 67 patients who received chemotherapy with or without thoracic radiation. Those authors observed that the 5-year OS rate was 27% in the surgical group versus 4% in the matched nonsurgical cohort. Brock et al16 reported on 82 patients who underwent surgery with neoadjuvant or adjuvant chemotherapy. They reported 5-year OS rates of 42% for the whole cohort and 58%, 18%, and 23% for patients with stage I, stage II, and stage III disease, respectively. A recent phase 2 trial17 examined the role of surgery in the platinum-based era by reporting on 62 patients who underwent surgery followed by 4 cycles of cisplatin plus etoposide. Their analysis revealed a favorable survival profile that included 5-year survival rates of 71% in patients with pathologic stage I disease, 38% in patients with pathologic stage II disease, and 39% in patients with pathologic stage IIIA disease.
An additional finding in the current study that requires comment is the use of PORT in these patients and its impact on survival. We observed that, for patients with N0 and N1 disease, there was no benefit from the addition of adjuvant radiation therapy, and there was a possible trend toward worse outcomes in patients with N1 disease. However, in patients with N2 disease, there was a significant survival benefit from the addition of PORT. To our knowledge, there are no other reports in the literature specifically analyzing the role of PORT in SCLC. However, in the NSCLC literature, there is similar evidence of a survival detriment from PORT in patients with N1 disease but a survival benefit23, 24, 26 in patients with N2 disease.
Limitations of this analysis include the lack of chemotherapy data on these patients, the inherent selection bias in retrospective reviews, the absence of central pathology review, the lack of some key pathologic variables (such as margin status), and the lack of information regarding performance status. In addition, we do not know why some patients were referred for PORT, because there may have been signs of more aggressive disease in those patients.
In an attempt to account the for lack of central pathology review, neuroendocrine and mixed NSCLC-SCLC tumors were not included in the current study. This was done with the intention of excluding tumors that may have represented carcinoid-type tumors or tumors with a significant NSCLC component, which may have better outcomes. In fact, we identified 606 patients with these histologies who did undergo surgery. The 5-year OS rate for this group was 47%, and the median survival was 54 months, which were significantly better than our reported 5-year OS rate of 34.1% and median survival of 28 months.
Because of the limitations of this study and the lack of a benefit from surgery after neoadjuvant chemotherapy in the previously mentioned LCSG trial,22 it is far too early to suggest that surgery should play a standard role in this disease. However, with only incremental improvements in survival to date using chemotherapy and/or radiation therapy, it is reasonable to suggest a review of the role of surgery on a prospective basis. For earlier stage disease, a strategy of definitive surgery followed by adjuvant platinum-based chemotherapy should be considered prospectively. For more locally advanced but still curable disease, options for prospective studies again may include definitive surgery followed by platinum-based chemotherapy. However, perhaps an aggressive strategy of trimodality therapy also should be considered that consists of preoperative chemoradiation followed by surgery and adjuvant chemotherapy. This approach has been used prospectively by the Essen Thoracic Oncology Group with a reported 5-year OS rate of 44% for those patients with stage IIB and IIIA disease and also currently is being studied in phase 3 randomized trials in Japan and Germany.27
In conclusion, in this population-based report, the use of surgery, particularly lobectomy, was associated with favorable survival outcomes, although these findings need to be confirmed on a prospective basis. However, this report does add to the growing literature suggesting that the role of surgery should be evaluated once again as part of the multimodality management of this disease.