Presented in part at the American Society of Clinical Oncology Annual Meeting; May 31-June 4, 2013; Chicago, IL.
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veteran Affairs.
Despite clinical trials demonstrating improved survival with adjuvant chemotherapy (AC) for patients with American Joint Committee on Cancer stages I to III non-small cell lung cancer (NSCLC), it is unclear whether this survival benefit extends to broader populations. The current study evaluated patterns of AC use and examined the impact of AC on survival.
A retrospective analysis was conducted of patients in the Veterans Affairs Central Cancer Registry diagnosed with stages IB to IIIA NSCLC between 2001 and 2008. Descriptive statistics were used to examine patterns of AC use over an 8-year time period. Cox proportional hazards regression analyses were used to estimate hazards ratios (HR) and 95% confidence intervals (95% CIs) to compare mortality risk among patients treated with and without AC.
Among 14,306 patients with stages IB to IIIA NSCLC, 4929 underwent surgery and 22% of these received AC. The percentages of patients diagnosed in 2001 through 2003, 2004 through 2005, and 2006 through 2008 receiving AC were 7.0%, 29.8%, and 29.5%, respectively. There was no survival benefit with AC noted for patients diagnosed between 2001 and 2003, but AC was associated with improved survival for the period between 2004 and 2005 (HR, 0.78; 95% CI, 0.67-0.91) and 2006 through 2008 (HR, 0.79; 95% CI, 0.69-0.91). Of those patients receiving AC, 89% received platinum-doublet chemotherapy. Carboplatin remained the most common agent, although cisplatin use reached 43% in the period between 2006 and 2008. The HR for cisplatin relative to carboplatin was 0.96 (95% CI, 0.80-1.15).
Despite receiving optimal surgery for localized NSCLC, greater than one-half of patients experience disease recurrence,[1, 2] and 5-year survival rates without additional treatment range from 67% for patients with stage IA disease to 23% for patients with stage IIIA disease. The goal of postoperative adjuvant chemotherapy (AC) is to reduce the risk of disease recurrence by eliminating residual disease that may persist after surgical resection, and thus improve survival. A landmark meta-analysis published in 1995 demonstrated a statistically nonsignificant trend in survival benefit associated with cisplatin-based AC compared with those patients treated with surgery alone, and an absolute survival benefit of 4% at 5 years. Subsequent randomized trials investigating the role of AC in patients with localized NSCLC reported conflicting results. Although some trials have found no survival difference between patients who underwent surgical resection with and without AC,[5, 6] 3 large trials demonstrated a statistically and clinically significant survival benefit with cisplatin-based AC regimens, with 5-year survival differences ranging from 4% to 15%.[7-9] Of these, the International Adjuvant Lung Cancer Trial was to our knowledge the largest and first positive trial demonstrating both an overall and disease-free survival benefit with AC. The Lung Adjuvant Cisplatin Evaluation (LACE) meta-analysis of these 5 large trials using individual patient data reported a significant benefit in overall survival with cisplatin-based AC (hazards ratio [HR], 0.89). The Cancer and Leukemia Group B trial, which examined the effect of the combination of carboplatin and paclitaxel for patients with resected stage IB disease, initially reported significantly better survival rates with AC at 4 years of follow-up in 2004, but subsequent analyses performed in 2006 did not reach statistical significance and were only suggestive of improved survival. A post hoc subset analysis suggested that tumors measuring > 4 cm had a survival advantage with this doublet. The LACE meta-analysis also found a nonsignificant trend toward prolonged survival among patients with stage IB disease treated with AC, although worse survival was observed for those with stage IA NSCLC. Therefore, survival benefit for patients with stage IA disease has not been proven and may be limited to patients with stage IB disease with a high risk of disease recurrence.
Evidence from randomized trials and meta-analyses performed within the last decade has had a great impact on clinical practice for nonmetastatic NSCLC. Currently, cisplatin-based doublet AC is the standard of care for patients with completely resected stage II and III disease.[1, 13, 14] However, the AC trials of the last decade included carefully selected patients with good physiologic function and the ability to tolerate cisplatin-based chemotherapy. In practice, many patients have significant comorbid disease burden and may have contraindications to cisplatin-based chemotherapy. Thus, evidence is lacking regarding the impact of AC on survival in large population-based cohorts. The current study evaluated the use of and outcomes associated with adjuvant chemotherapy among patients after surgical resection for stage IB to IIIA NSCLC.
MATERIALS AND METHODS
Study Population and Data Sources
This population-based retrospective study used electronic data within the Veterans Affairs (VA) health care system and was approved by the Durham VA Institutional Review Board. The study population consisted of patients diagnosed with American Joint Committee on Cancer (AJCC) stages IB to IIIA NSCLC between January 1, 2001 and December 31, 2008 who underwent surgical resection. Data for this study were obtained from the national VA Central Cancer Registry and the VA Corporate Data Warehouse. The VA Central Cancer Registry contains all patients diagnosed with and/or treated for cancer in the VA and was used to identify the cohort of patients diagnosed with stages IB, II, and IIIA NSCLC between 2001 and 2008 who underwent surgery and provide data regarding patient demographics, tumor characteristics, and primary treatment. Registry data were linked to several databases in the VA Corporate Data Warehouse databases to obtain pharmacy, inpatient and outpatient diagnoses, and vital status data.
Exposures and Outcomes
The primary exposure was use of AC, defined as the administration of chemotherapy drugs within 4 months after surgical resection. Pharmacy data from the Decision Support System databases were used to identify patients receiving chemotherapy, including the drug name and date of administration. A dichotomized variable was created for AC use. Clinical and demographic factors assessed included patients' age at diagnosis, race (white, black, other/unknown), diagnosis period (2001-2003, 2004-2005, and 2006-2008), American Joint Committee on Cancer stage at diagnosis (stages IB, II, and IIIA), marital status (unmarried, married, and unknown), and comorbidity. We obtained information regarding preexisting conditions from inpatient and outpatient administrative data. The Charlson comorbidity index was used to classify these comorbidities and consists of 19 conditions determined by International Classification of Diseases, Ninth Revision (ICD-9) codes. We assessed diagnosis of each condition documented within 18 months before the lung cancer diagnosis date and further classified patients based on whether their comorbidity index was 0, 1, or 2, or ≥ 3. Overall survival was the primary outcome. Mortality was determined by the date of death in the VA Vital Status File (the accuracy of vital status from this source is considered comparable to the National Death Index). Mortality was assessed from the date of diagnosis through July 1, 2013 (censor date).
Descriptive statistics were used to examine differences in demographic and clinical characteristics according to the time period of diagnosis. Overall survival was calculated as years between the date of diagnosis and either the date of death or censoring. Cox proportional hazards modeling was used to estimate the prognostic significance of demographic and clinical characteristics. Multivariate models were adjusted for age, race, marital status, comorbidity, AJCC stage of disease, and time period. The 2 treatment comparisons were AC versus no AC, and carboplatin-based versus cisplatin-based AC for the subset of patients receiving AC. Overall survival curves were estimated using the Kaplan-Meier method. All significance tests were 2-sided and P values < .05 were considered statistically significant. Analyses were conducted using SAS statistical software (version 9.2; SAS Institute Inc, Cary, NC).
A total of 48,844 patients were diagnosed with stages I to IV NSCLC between 2001 and 2008. Figure 1 illustrates the percentage of patients who were diagnosed, underwent surgical resection, and survived 4 years after diagnosis (Fig. 1). There was no evidence of a stage shift over the 8-year time frame (eg, from increased use of positron emission tomography scans) (Fig. 1A). Resection rates were relatively stable over time, with the most fluctuation observed for patients with stage II disease (Fig. 1B). Figure 1C shows no improvement in 4-year survival rates among patients with stage IV disease, further suggesting no significant stage shift over time; there were small but steady increases in survival among patients with stages I to III disease.
Among 14,306 patients diagnosed with stages IB to IIIA NSCLC between 2001 and 2008, 5967 underwent surgical resection. The following sequential exclusions were applied: unknown surgery date (11 patients), neoadjuvant treatment (264 patients), death ≤ 30 days after surgery (243 patients), and adjuvant radiotherapy (520 patients). These exclusions resulted in a final study population of 4929 surgically resected patients. Most characteristics of the study population were consistent over the 3 time periods (Table 1). The median age was 67 years, a majority of the patient population (84%) was white, and 20% had no comorbidities. Overall, 22% of patients in the sample received AC, with a dramatic increase from 7% between 2001 and 2003 to approximately 30% in subsequent time periods. There was a nearly 7% increase in the 4-year survival rate of the study population between 2001 and 2003 (43%) and 2006 and 2008 (50%).
Table 1. Characteristics Of Patients Who Underwent Surgical Resection for Stage IB to IIIA NSCLC by Time Period of Diagnosis
All Years (N = 4929)
2001–2003 (N = 1611)
2004–2005 (N = 1311)
2006–2008 (N = 2007)
Abbreviation: AJCC, American Joint Commission on Cancer; NSCLC, non-small cell lung cancer.
For all years combined, the HR estimate for age was 1.02 (95% confidence interval [95% CI], 1.02-1.03) (Table 2). Race was found to have no statistically significant association with survival, married status correlated with improved survival, and patients with stages II and III disease were found to have worse survival compared with patients with stage IB disease. Compared with having no comorbidities, only the association with ≥ 3 comorbidities was found to be statistically significant (HR, 1.29; 95% CI, 1.17-1.42). AC was associated with an 18% improvement in survival for all years combined (HR, 0.82; 95% CI, 0.74-0.90).
Table 2. HRsa and 95% CIs for the Association Between Demographic and Clinical Characteristics and OS by Time Period of Diagnosis
Abbreviations: 95% CI, 95% confidence interval; AJCC, American Joint Committee on Cancer; HR, hazards ratio; OS, overall survival.
Adjusted for age, race, marital status, AJCC stage, comorbidity, and receipt of adjuvant chemotherapy and for all years combined, time period.
In analyses stratified by time period of diagnosis, increasing age, AJCC stage, and comorbidity were associated with worse survival in each time period (Table 2). HRs were significantly greater for patients with advanced stages of disease compared with those with stage IB disease and this was the case for each time period. All HR estimates for ≥ 3 comorbidities were statistically significant and varied little over time (HR range, 1.27-1.32). AC was not associated with survival for the period between 2001 and 2003, but correlated with significant survival benefit for 2004 through 2005 (HR, 0.78; 95% CI, 0.67-0.91) and 2006 through 2008 (HR, 0.79; 95% CI, 0.69-0.91). Race was not associated with survival, and married status demonstrated improved survival only for 2004 through 2005. In preliminary analyses, only age, AJCC stage, and marital status were found to be significantly associated with receipt of AC for all years combined (see online supporting information).
Stage-Specific Patterns of Adjuvant Chemotherapy and Overall Survival
In all years combined, AC was associated with improved survival in each stage of disease (Table 3), with the strongest benefit observed for stage II (HR, 0.77; 95% CI, 0.66-0.89). When stratified by time period, only stage II had a statistically significant association with survival for 2001 through 2003 and 2004 through 2005. AC demonstrated reduced mortality risk for diagnoses of stage IB disease (HR, 0.73; 95% CI, 0.58-0.93) and stage IIIA disease (HR, 0.73; 95% CI, 0.54-0.98) between 2006 and 2008. In all stages and time periods combined, the 4-year overall survival rate among patients treated with AC was 52% compared with 46% for those treated without AC (data not shown).
Table 3. HRsa and 95% CIs for the Association Between Adjuvant Chemotherapy and OS, by Stage
Adjusted for age, race, marital status, and comorbidity and for all years combined, time period.
Patients receiving AC were primarily treated with multiple agents, most commonly platinum-based doublet AC (89%) (Table 4). Carboplatin remained the most frequently used platinum agent; however, among patients treated with platinum-based doublet AC, the administration of carboplatin increased from 73% (74 of 101 patients) in 2001 through 2003 to 90% (325 of 362 patients) in 2004 through 2005 and then decreased to 57% (296 of 517 patients) in 2006 through 2008. Cisplatin use decreased from 27% in from 2001 through 2003 to 10% in 2004 through 2005 and increased to 43% in 2006 through 2008. Thus, there was an overall 16% increase in cisplatin-based doublet AC between 2001 and 2003 and 2006 and 2008. The combination of carboplatin and paclitaxel was the predominant carboplatin-based combination used for all time periods. The combination of cisplatin and etoposide was the most common cisplatin-based doublet in 2001 through 2003 and 2004 through 2005, but the combination of cisplatin and vinorelbine was the primary pair in 2006 through 2008. Cisplatin and carboplatin AC had similar survival outcomes in all patients, and in all time periods, the adjusted and unadjusted HRs for cisplatin were not statistically significant (Fig. 2).
Clinical practice guidelines appear to favor cisplatin-based AC for patients with stages II and III resected NSCLC based on evidence from clinical trials. Population-based studies are necessary to determine whether outcomes observed in clinical trials produce similar results in the general oncology population. We believe the current study, a large population-based study performed among veterans, provides a real-world evaluation of AC use between 2001 and 2008, the era of its adoption in the United States, and its impact on overall survival. The results of the current study demonstrated a significant survival benefit associated with receipt of AC. The magnitude of the benefit in the current study (HR, 0.82) is comparable to that observed in clinical studies (HR, 0.89). We further noted that this association varied by AJCC stage and time period of diagnosis. Carboplatin was the primary platinum agent used in the adjuvant setting and the combination of carboplatin and paclitaxel was the most common doublet. However, cisplatin-based chemotherapy was associated with similar overall survival rate compared with carboplatin-based chemotherapy and its use increased in the period 2006 through 2008 after the final results of what to our knowledge is the only clinical trial using carboplatin-based chemotherapy published to date.
We observed a dramatic increase in receipt of AC between 2001 and 2003, when there was little evidence regarding the effect of AC, and 2004 and 2005, when randomized trials reporting a survival advantage with chemotherapy began to emerge. To our knowledge to date, the only published large population-based study of AC uptake for NSCLC was performed in a population of surgical patients in Ontario, Canada. The authors noted an increase in the adoption of AC since 2004 when initial trials were published. They reported AC rates of 7% in 2001 through 2003 and 31% in 2004 through 2006, rates that are very similar to the rates of AC use in the current study. These rates may be underestimated due to our use of registry and administrative data, which do not capture the reasons patients may not be eligible for AC. Ryoo et al conducted a study among US veterans diagnosed with NSCLC in 2007 and found that slightly less than one-half of eligible patients received adjuvant chemotherapy for resected stage II/III disease. Other studies of AC uptake after 2004 reported use rates of approximately 40% among eligible patients at single facilities.[21, 22]
When evaluating the effect of AC on outcome over time, it is important to evaluate possible changes in diagnosis, staging, and treatment patterns. We considered, but did not observe, variation in surgical treatment patterns and evidence of a stage shift between 2001 and 2008 that would significantly influence survival outcomes. Other clinical factors such as stage of disease and preexisting conditions are prognostic factors for NSCLC and could affect the correlation between treatment and survival. Pathologic stage is the main prognostic factor guiding recommendation of AC and comorbid conditions can affect patients' response to treatment as well as their likelihood of completing treatment. In the current study sample, increasing stage of disease was positively associated with receipt of AC, but comorbidity was not. When controlling for these factors, we found a significant survival benefit with AC, particularly among patients diagnosed between 2004 and 2005 and 2006 and 2008. Based on the LACE meta-analysis, we would expect an absolute 5-year survival benefit of 5%. Our finding of a 6% absolute improvement in 4-year survival (52% with AC and 46% without AC) was within the range of that observed in other studies.[8, 10, 19] Our strongest association with survival was observed for diagnoses of stage II disease made between 2001 and 2003, which has been the focus of most trials. It is possible that the significant survival benefit with AC noted among patients with stage IB disease diagnosed between 2006 and 2008 was influenced by the overall improvement in survival for patients with stage I disease relevant to previous time periods. However, the management of stage IB disease with AC remains controversial. Several trials have failed to demonstrate a significant survival advantage for patients with stage IB disease, but subsequent analyses have suggested that survival advantage may be limited to patients with stage IB disease whose tumors measure > 4cm.[8, 11]
As expected, a majority of patients in the current study received platinum-based AC. Carboplatin was the most common platinum agent. Although a great deal of patient selection is involved when recommending carboplatin versus cisplatin regimens in clinical practice, current guidelines and available evidence from trials most strongly support cisplatin as the optimal platinum agent in the adjuvant setting if it can be tolerated. In what to our knowledge is the only adjuvant study to date using carboplatin was ultimately reported as not demonstrating a significant survival advantage, but this study was underpowered. We observed a similar overall survival with cisplatin-based AC compared with carboplatin. Gu et al recently reported the more frequent use of carboplatin as well as similar survival between carboplatin-based and cisplatin-based AC among patients in the Surveillance, Epidemiology, and End Results-Medicare database who were diagnosed with NSCLC between 1992 and 2007. Although randomized controlled trial data are a higher level of evidence for assessing treatment effectiveness, population data have indicated that carboplatin-based AC is as effective as cisplatin-based AC.
The nonplatinum agent paired with cisplatin or carboplatin may contribute to the survival benefit associated with AC; however, to the best of our knowledge, there is little definitive evidence regarding which nonplatinum drug is best to pair with the platinum agent. The predominant regimen used in clinical trials is cisplatin and vinorelbine, suggesting that it may be the preferred doublet; however, this could also reflect the availability of drugs during the time of the trials or a bias of those designing and funding the larger adjuvant trials. In the current study, the combination of cisplatin and vinorelbine became the primary cisplatin doublet in the period between 2006 and 2008. Data regarding the efficacy of carboplatin doublet AC stems from the underpowered Cancer and Leukemia Group B trial, in which the combination of carboplatin and paclitaxel did not demonstrate a survival benefit for patients with stage IB disease, which is contrary to our finding of improved survival with AC among patients with stage IB disease. Ongoing phase 3 trials[27-29] are currently assessing the efficacy of adding bevacizumab to cisplatin-doublet chemotherapy (Eastern Cooperative Oncology Group E1505; ClinicalTrials.gov identifier NCT00324805[27, 28]), biomarker-guided selection of chemotherapy (TASTE [TAilored Post-Surgical Therapy in Early Stage NSCLC] [ClinicalTrials.gov identifier NCT00775385[27, 28]], GECP-SCAT [ClinicalTrials.gov identifier NCT00478699[27, 28]] ITACA), use of an epidermal growth factor receptor inhibitor based on tumor epidermal growth factor receptor status (RADIANT [A Study of Tarceva After Surgery With or Without Adjuvant Chemotherapy in Non-Small Cell Lung Carcinoma (NSCLC) Patients Who Have Epidermal Growth Factor Receptor (EGFR) Positive Tumors] [ClinicalTrials.gov identifier NCT00373425[27, 28]]; EMERGING [Erlotinib Versus Gemcitabine/Cisplatin as (Neo)Adjuvant Treatment in Non-small Cell Lung Cancer] [ClinicalTrials.gov identifier NCT01407822]; ADJUVANT [ClinicalTrials.gov identifier NCT01405079]; and ICTAN [Icotinib Following Chemotherapy Versus Chemotherapy as Adjuvant Therapy in Stage IIA-IIIA NSCLC With EGFR Mutation] [ClinicalTrials.gov identifier NCT01996098]), and immunotherapy (MAGRIT; ClinicalTrials.gov identifier NCT00480025). These studies include chemotherapy regimens not commonly used in the current study, including cisplatin doublets with docetaxel, gemcitabine, and pemetrexed, and single-agent paclitaxel, docetaxel, and pemetrexed. Results from these studies may provide insight regarding the optimal drug combination and change current guidelines and practice patterns for AC in patients with resected NSCLC, as did the pivotal trials indicating a survival benefit with AC.
A major advantage to the current study was its large sample size, including all patients with NSCLC diagnosed or treated in the VA health care system, from which to assess use of AC. Another advantage was the ability to assess temporal changes in AC use, particularly time periods before, during, and after the publication of pivotal randomized trials of AC for NSCLC. This study improves on other population-based studies by our ability to assess the extent to which stage modifies the association between AC and survival. However, there are limitations that should be noted. Data sources for the current study did not include patients' performance status, which is an important prognostic factor for consideration in recommending AC. Patients most likely to benefit from AC are those in generally good health and who are relatively young, without significant comorbidities and no postoperative complications. In an effort to minimize the influence of patients in poor health or those with more severe disease and therefore a worse prognosis, we excluded those patients who died within 30 days after surgery and those who received radiotherapy. We were not able to capture AC use outside of the VA or examine tumor size within stage IB for consideration in these analyses. In addition, given the nature of the VA health care system, issues regarding access to care are minimized and may have resulted in higher rates of AC use and completion as well as follow-up care, limiting the generalizability of the current study results to other populations.
There was a substantial increase in AC use noted between 2004 and 2005; however, a majority of patients still do not appear to receive AC and reasons for this nonreceipt of AC need to be explored further. AC use was associated with significant survival advantage for patients diagnosed between 2004 and 2005 and 2006 and 2008. Although associations with improved survival varied by stage of disease, there was no difference noted with regard to overall survival when comparing carboplatin-based and cisplatin-based regimens. If cisplatin tolerance is deemed to be a limiting factor in receipt of AC, then the finding that carboplatin-based chemotherapy may offer a similar benefit is particularly relevant. Together, these findings indicate that clinical trial results regarding adjuvant chemotherapy benefit are applicable in less selective populations and support the use of AC in routine clinical practice. This highlights the importance of identifying patients who are most likely to benefit from adjuvant chemotherapy and improving AC use rates in an effort to improve patient outcomes.