Adjuvant Chemotherapy for Resected Non-Small Cell Carcinoma of the Lung: Why We Still Don't Know

Abstract

The role of adjuvant chemotherapy in resected non-small cell carcinoma of the lung (NSCLC) remains controversial. A critical review of the 11 cisplatin-based randomized trials addressing this issue was performed using methodology adapted from the CONSORT statement. The 11 trials were divided into those that included predominantly node-negative patients (n = 5) and those that included predominantly node-positive patients (n = 6). In the node-negative trials, which included 1,084 evaluable patients, no evidence of heterogeneity of treatment effect between the trials was found. The composite five-year survival rate for chemotherapy patients was 61% versus 55% for control patients, which reached marginal significance (p = 0.06). In the node-positive trials, which included 880 evaluable patients, a marginal lack of homogeneity of treatment effect between trials was found (p = .013). The composite two-year survival rate for chemotherapy patients was 48% versus 40% for control patients, which reached marginal significance (p = 0.06). Although not definitive, these trials suggest a benefit to postoperative adjuvant chemotherapy in resected NSCLC. The cumulative experience of the 11 trials is notable for: A) the small number of patients; B) heterogenous patient populations; C) substandard chemotherapy regimens; D) inadequate chemotherapy delivery, and E) less than optimally executed clinical trials. The 11 trials are discussed in detail, and recently completed and ongoing trials of adjuvant chemotherapy in resected NSCLC are reviewed.

Introduction

Non-small cell lung cancer (NSCLC) remains the leading cause of cancer death in both men and women [1]. Surgical resection in the early stages of disease results in long-term survival in the majority of patients. However, up to 30% of Stage I and 50% of Stage II patients will die despite surgery [2, 3]. The postoperative prognosis depends on such factors as T stage of the primary tumor and nodal status [4]. The vast majority of relapses following a curative resection are systemic in nature, with local recurrence representing only a minority of cases [5, 6]. Adjuvant radiotherapy has been shown to reduce the incidence of local recurrences; however, it does not improve survival [7]. Given these findings, the strategy that would have the greatest impact on mortality is effective postoperative chemotherapy that significantly reduces the occurrence of systemic relapse.

Although postoperative adjuvant chemotherapy improves survival in several solid tumors, its role in NSCLC remains controversial [8]. A recent meta-analysis suggests that early trials which employed alkylating agents in the postoperative setting have a detrimental effect on survival [9]. In the cisplatin era, 11 randomized trials have been reported [10-20]. The patient populations studied in these trials have varied from Stage I to Stage III, and many of the trials were designed and carried out in the early to mid-1980s, employing chemotherapy regimens and supportive care measures reflective of that time period. The recent meta-analysis [9] has suggested a reduction in the odds ratio for death in those trials employing postoperative cisplatin-based adjuvant chemotherapy. However, it is generally believed in the oncologic community that adjuvant chemotherapy has no role in the management of curatively resected NSCLC [21]. We believe that a lack of sufficient data exists to fully resolve this question rather than that the existing data sufficiently exclude a positive role of adjuvant chemotherapy in curatively resected NSCLC. After all, absence of proof is not proof of absence. The purpose of this review is to describe and critically analyze the published and ongoing trials of adjuvant cisplatin-based chemotherapy in resected NSCLC.

Methods

The studies included in this review were obtained by searching the English language literature in Medline from 1980 to the present. The following keywords were entered: non-small cell lung cancer, adjuvant chemotherapy, randomized trial. Only completed and published randomized trials using platinum-based regimens after attempted curative resection of non-small cell lung cancer were included in the analysis. Additional studies were obtained from reference lists of the above papers, review articles, and the authors' review of pertinent medical journals. Abstracts were not included since they lacked sufficient detail for careful analysis.

Eleven studies were identified. Each study was examined for the number of patients enrolled, balance of control and chemotherapy arms with respect to prognostic factors, statistical methods, statistical significance, chemotherapy regimen employed, compliance with treatment regimens, reasons for lack of compliance, toxicity and overall survival.

A methodologic scoring system was developed to evaluate the quality of the design and to report the results of the trials. The criteria used were adapted from the CONSORT statement, which applies to randomized clinical trials focusing on survival data [22]. The 13 criteria selected by the four authors (Table 1) were felt to be the essential criteria with which a study group should design, execute, and report a clinical trial addressing the role of adjuvant chemotherapy in NSCLC. Furthermore, these criteria were felt to be essential to providing sufficient information to enable a reader to draw meaningful conclusions from the data. The authors independently scored each criterion using a scale of 0, 0.5, and 1.0: 0 if the criterion was clearly not met by the study, 0.5 if the criterion was partially met by the study, or 1.0 if the criterion was clearly met by the study. The authors include a medical oncologist (MAS), a surgical oncologist (FCD), a biostatistician (MJS), and an internal medicine resident (SG). Average scores were determined by averaging the individual scores of the four reviewers with thirteen, indicating a perfect score.

For this review, the trials were divided into those that included predominantly node-negative patients (defined as ≥60% node-negative patients entered into the trial) and those that included predominantly node-positive patients (defined as ≥60% node-positive patients entered into the trial).

Completeness and Quality of Reporting

Many of the trials did not include sufficient data from which to draw meaningful clinical conclusions. As noted above, a scale (Table 1) modified from the CONSORT statement [22] was used to objectify the completeness and quality of reporting in each of the 11 trials. Table 2 shows the average score for each of the 11 studies reviewed. Overall, there was good agreement among the authors regarding study reporting quality. Correlations between authors ranged from .55 to .95, the latter representing the strong agreement between MAS and FCD. The correlation between each author and the average score ranged from .80 to .98. For 68% of the individual criteria, three of the four authors issued the same score. The remaining cases primarily represent more lenient versus tougher scoring by the reviewer, which was consistent for an individual criterion. Overall, the studies adequately addressed criteria 1, 2, 4, and 12 with average scores of .86, .93, .75, and .78, respectively. No prospectively defined stopping rules were used for any of the studies, so all authors gave zero scores for criterion 5. The average scores for the remaining eight criteria ranged from .38 to .56. Although there were some gaps in methods and reporting, an attempt was made to combine the study results to determine if a trend exists for the effect of chemotherapy for either predominantly node-negative or node-positive patients.

Statistical Methods

An overall assessment of the treatment effect from the studies examined in this article was made using the random effects model of DerSimonian and Laird [23]. A test for homogeneity of the relative odds of survival across the studies was done using the statistic Q_w from that article. The survival rates at two and five years were either obtained directly from the study or were estimated from the Kaplan-Meier curves. Unfortunately, sufficient information regarding the censoring of patients who did not die was not provided in the papers. Thus, we assumed that all censoring occurred after the time point of interest. While this is not likely to be true, this provides us with an overall strength of the combined data if the survival rate estimates are accurate and full follow-up did occur. As such, it provides us with the greatest possible information from the different trials for the sample sizes used.

Predominantly Node-Negative Trials

Overall, 1,084 evaluable patients were studied in the five predominantly node-negative trials (Table 3). The composite five-year survival rate of the chemotherapy arm patients (n = 598) was 61%. The composite five-year survival rate of the control arm patients (n = 486) was 55%. When examined individually, the treatment effects at five years for the five predominantly node-negative trials ranged from a 13% benefit for chemotherapy [10] to a 1% benefit for surgery alone [14]. There was no evidence of heterogeneity of effect between the trials (Q_w =1.68, p = .79). Overall, a 5.6% marginally significant five-year survival advantage for chemotherapy was obtained, with the standard error being 3.0% (p = .06). Each of these five trials is summarized briefly below.

Wada et al. [10] randomized 323 completely resected NSCLC patients to either cisplatin, vindesine, Uft (CVUft), Uft alone, or observation (see Table 3 for dosing schedule). Three hundred ten patients were eligible for analysis. Follow-up was 100% at five years. The three groups were well matched in terms of sex, age, histological classification, pathologic T and N status, and pathologic stages. The Uft group experienced a significant survival advantage compared to the control group (p = .02). A multivariate analysis with Cox's proportional hazards model identified postoperative adjuvant chemotherapy as a significant prognostic factor for survival, with hazards ratios (against the control group) of 0.64 (95% CI, 0.42 to 0.97) in the CVUft group and 0.55 (95% CI, 0.36 to 0.86) in the Uft group. Specific data regarding the need for dose reduction or discontinuation of treatment secondary to toxicity were not presented.

In 1995, the Study Group of Adjuvant Chemotherapy for Lung Cancer in Japan published a trial [11] with a similar chemotherapy arm as in the Wada trial [10]. This study compared cisplatin, adriamycin, and Uft (CAUft) versus observation alone following curative resection in 309 patients. Compliance in this study was not well reported. It was noted that 14.9% of patients had chemotherapy “interrupted” at some point, although specific data regarding the reasons for interruption and mean total dose received were not reported. The trial did not find a statistically significant survival difference in the chemotherapy group versus control (p = 0.35). However, patients enrolled in this trial were not stratified for prognostic factors and the CAUft arm had significantly more advanced pathological lymph node stage than the control group. When the authors adjusted for this variable, CAUft did confer a significant survival advantage at five years when compared to controls (p = 0.044).

In 1993 Feld et al. [12] reported a trial comparing cyclophosphamide, adriamycin, and cisplatin (CAP) versus observation alone following complete resection of T₁N₁ or T₂ N₀ disease (Table 3). No difference in survival between the CAP and observation arms was observed. Compliance in this study was poor, with only 53% of patients receiving all four planned cycles of chemotherapy and only 57% of this subgroup receiving chemotherapy on time. A full 20% of patients in the chemotherapy arm received no treatment at all. Reasons for the poor compliance were not directly discussed, although it was reported that 88% of patients had mild to severe nausea and vomiting and 65% had hematologic toxicity. Overall, no survival difference was found (p = .92). However, recurrence and death rates in the squamous cell population both favored the chemotherapy arm (p = 0.05), while the non-squamous cell group fared less well than the control group (p = 0.06).

In 1992, Niiranen et al. [13] published results on 110 patients randomized to receive either adjuvant CAP or observation following surgery. Fifty-seven percent of the patients on the CAP arm received the planned therapy, with 70% of patients receiving at least three cycles. The timing of the chemotherapy received relative to the planned schedule was not reported. Thirty percent of patients in the CAP arm refused treatment at some point and 13% received no chemotherapy at all (22% of patients refused because of nausea and vomiting). The CAP arm had a better survival (p = 0.05) with a five-year survival rate of 67% in the CAP arm versus 56% in the control arm. The groups were well balanced in terms of prognostic factors, including pathological T and N stage, gender, and histological subtype. However, more patients underwent pneumonectomy in the control population (22 versus 11 patients). Patients who underwent pneumonectomy had a significantly worse outcome than patients undergoing lobectomy (p = 0.002), representing a possible bias against the control group. The five-year survival rate for those patients undergoing lobectomy was 74% in the CAP group and 64% in the control group (p = 0.24).

Ichinose et al. [14] published a trial in 1991 comparing three cisplatin-based chemotherapy arms with observation following curative resection. The chemotherapy arms included CAP with mitomycin, cisplatin-vindesine, and cisplatin-adriamycin-pepleomycin (Table 3). The authors' goal was to deliver “at least two cycles” of treatment. Compliance was reported as “good” for all patients. The only specific compliance data reported was that 30% of patients refused treatment beyond two cycles, and an average of only 2.3 cycles was delivered (range 1-6). No significant survival difference between the arms was seen.

Predominantly Node-Positive Trials

Overall, 880 evaluable patients were studied on one of six predominantly node-positive trials (Table 4). The composite two-year survival rates of the chemotherapy and control arm patients were 48% (n = 437) and 40% (n = 443), respectively. Ohta et al. [15] and Kimura et al. [16] both had notably higher survival rates in the chemotherapy arms, while Ohta et al. [15] also had an unusually high survival rate in the control arm compared with the other trials. Excluding those two trials, the two-year survival rates drop to 39% and 33%, and the remaining trials appeared statistically homogeneous. The treatment effects for the six predominantly node-positive trials ranged from a 31% advantage for chemotherapy [16] to a 13% benefit for surgery alone [17]. A marginal lack of homogeneity of treatment effect was found (Q_w = 12.03, p = .13). Besides the two extreme treatment effects (which also had the smallest sample sizes), the effects of the four other trials clustered between a 4% and 11% benefit for chemotherapy. Overall, an 8.2% two-year survival advantage for chemotherapy was obtained, which is marginally significant (p = .06), since the standard error is 4.3%. In general, these trials were small, had poor compliance with treatment and heterogeneous patient populations and often did not reach their accrual goal. It is therefore difficult to interpret these studies, each of which is briefly summarized below.

Dautzenburg et al. [18] reported a randomized trial of 267 patients who underwent complete resections. Patients were randomized to adjuvant radiotherapy (60 Gy in six weeks) or three courses of cyclophosphamide, doxorubicin, cisplatin, vincristine, and lomustine followed by the same course of radiotherapy (Table 4). Compliance with treatment was reportedly good, with 70% of patients in the radiotherapy arm receiving greater than 50 Gy, 88% of chemotherapy patients receiving greater than two-thirds of the planned drugs, and 67% of these patients receiving at least 50 Gy. Specific reasons for noncompliance were not addressed, although 8% of patients reportedly had grade 3-4 toxicity, 6% with grade 3-4 hematologic toxicity, and three patients reportedly died from the complications of chemotherapy. This study found no significant difference in survival (p = 0.63). In a subset analysis, patients with N₂ disease treated with chemotherapy plus radiotherapy had a significantly better survival than the radiotherapy-alone group (p = 0.003) with five-year rates of 19% and 6%, respectively. This survival advantage appeared to be due to a reduction in distant metastases in the N₂ patients receiving chemotherapy (44% versus 18% at five years, p = 0.002). There was no difference in local relapse. In the N₀/N₁ group, however, a reverse trend was observed, with poorer survival in those patients treated with chemotherapy (p = 0.03). The authors postulate that the reason for this discrepancy may be that N₂ patients have higher metastatic potential and therefore may gain more from systemic treatment. This trial did not meet its accrual goal. The study was designed to show a 10% survival increase in the chemotherapy group at one year with a type I error of 0.05 and 90% power. Given these assumptions, 362 deaths were required using a one-sided test. Although a four-year accrual time was planned, the study was closed after five years with only 233 deaths.

A study by Pisters et al. [17] in 1994 reported a randomized trial of radiotherapy plus or minus vindesine and cisplatin in 72 N₂ patients having undergone surgical resection. Sixty-one percent of the patients had a complete resection, while 39% were incompletely resected. (No definition for an “incomplete resection” was given). All patients received mediastinal radiation (40 Gy), and patients with incomplete resections received intraoperative brachytherapy (¹²⁵I implants). The groups were well balanced for T stage, histology, performance status, and extent of resection. Thirty-six patients were randomized to receive chemotherapy. Only 13 patients (36%) received all four doses of cisplatin. Ninety-two percent of the patients received the planned dose of external beam radiation, and 71% of eligible incompletely resected patients received implants. Reasons for noncompliance were not reported in detail. No significant survival advantage was found for either arm. This trial was initially designed to enroll 50 patients per arm; however, it was stopped early because of “noncompliance with the randomization process.” The study had little power to identify either a benefit or a detriment to chemotherapy in this setting.

Ohta et al. [15] in 1993 reported a randomized trial of 209 Stage III patients receiving either postoperative cisplatin and vindesine or observation alone, 181 of whom were analyzed. All patients underwent a complete surgical resection. The patients were well balanced between treatment arms except for age, with significantly older patients in the control arm (56 versus 59 years, p = 0.05). There were fewer patients with N₂ disease and more patients with T₃N_0/1 in the control arm than in the chemotherapy arm. Although this tended to benefit the control arm, these differences did not achieve statistical significance. Only 41% of patients received all three cycles of chemotherapy as planned. The average cumulative dose was reported to be 68% of the planned dose. The most frequent reason for patients not receiving the second cycle was patient refusal (occurring in 43% of cases). Survival did not significantly differ between the two groups (p = 0.86). A multivariate analysis showed that advanced pathologic N stage was associated with a poor outcome. Survival for pN₀ was 62%, pN₁ 44% and pN₂ 29% (p = 0.01). Given this finding, it is somewhat disconcerting that there were more patients with N₂ disease in the chemotherapy arm. Also, this trial was designed to detect a 20% increase in the three-year disease-free survival in patients receiving postoperative chemotherapy but was stopped slightly short of the planned accrual of 100 evaluable patients in each group.

Lad et al. [19] published a trial of 172 patients with locally advanced, incompletely resected NSCLC randomized to receive radiation therapy (40 Gy) plus or minus six cycles of CAP. One hundred sixty-four patients were eligible for analysis. There were significantly more N₁ patients in the chemotherapy arm than in the control (20 versus 10, p = 0.03). Only 51% of patients received all six cycles. Reasons for noncompliance were not provided. The survival rate at one year was 68% for the chemotherapy arm versus 54% for the radiation arm (p = 0.02); however, the death rates did not differ after the first year. The study examined both recurrence and death rates on a rate per person per year and found that the chemotherapy group had significantly lower recurrence and death rates per person than the radiation alone arm during the first year.

Holmes et al. [20] randomized 141 patients with either Stage II or III adenocarcinoma or large-cell undifferentiated carcinoma to either four cycles of CAP chemotherapy or intrapleural BCG/levamisole/isoniazid. One hundred thirty patients were eligible for analysis. The survival at two years was 41% for the chemotherapy arm versus 30% in the immunotherapy arm (p = 0.08). Compliance in the chemotherapy arm was poor, with 24% of patients receiving no chemotherapy at all. The reported average cumulative dose was 58% of the planned dose. Reasons for noncompliance were not specified, although 87% of patients suffered gastrointestinal toxicity and 48% hematologic toxicity. The two groups were reportedly well balanced for prognostic factors; however, the control arm had significantly more T₃ disease than the treatment arm (p = 0.024). Adjustment for T status as well as stage and performance status did not significantly alter their results.

Kimura et al. [16] reported a small series of adjuvant immunochemotherapy following complete resection of Stage II and III patients with NSCLC. A total of 82 patients were enrolled according to the treatment arms in Table 4. The study was originally designed as a three-arm study: observation, chemotherapy alone, and immunochemotherapy. In the first three years of the study, the authors stated that the results in the chemotherapy alone arm were so “poor” that this arm was subsequently discontinued, and randomization continued only between observation and immunochemotherapy. Over the course of the study, 33 patients were randomized to immunochemotherapy, 13 patients to chemotherapy alone, and 36 patients were in the observation arm. Overall survival was reported to be significantly better in the immunochemotherapy arm than in the controls, (p = 0.004). Patients who died without a recurrence were to be censored cases. Specific data concerning the number of censored cases were not given. All patients in the immunochemotherapy group received at least two cycles of chemotherapy and two cycles of LAK, IL- 2.

Why We Still Don't Know

The role of adjuvant cisplatin-based chemotherapy in resected NSCLC remains uncertain despite the trials discussed above that were designed to address the issue. In general, all the trials have similar problems: A) small numbers of patients limiting the statistical power of the study; B) heterogenous patient populations; C) substandard chemotherapy regimens by current-day standards; D) inadequate delivery of chemotherapy, likely secondary to the difficulty in delivering postoperative chemotherapy in this population and substandard supportive care by current-day standards, and E) less than optimally executed clinical trials. Most of the trials enrolled heterogenous patient populations ranging from Stage I to IIIA. This combines groups of patients with dramatically different survival outcomes [24]. Although this results in more rapid accrual, it yields subpopulations among which overall prognosis varies dramatically. Despite these problems, our analysis suggests a trend reaching marginal significance toward improved survival in those patients receiving adjuvant chemotherapy after resection for NSCLC.

In the recent meta-analysis [9], trials were divided between those that included either of the following:

In the former group, 1,394 patients were analyzed and the overall hazard ratio favored chemotherapy (0.87), resulting in a reduction in the risk of death of 13%. The absolute benefit, which was 3% at two years and 5% at five years, was not statistically significant. In the latter group, 668 patients were analyzed and the overall hazard ratio again favored the chemotherapy group (0.94), resulting in a reduction in the risk of death of 6%. Again, the absolute benefit, which was 2% at two years and 5% at five years, was not statistically significant.

The small number of patients (2,062) included in the NSCLC meta-analysis limits the statistical power of that analysis. In contrast, the meta-analysis examining the effect of adjuvant chemotherapy in breast cancer had approximately 75,000 patients involving 31,000 recurrences and 24,000 deaths [25]. In the breast cancer meta-analysis, a 6% improvement in 10-year survival was demonstrated for chemotherapy. Individually, most of the trials in either breast or colon cancer that have shown a benefit to adjuvant chemotherapy have included substantially more patients than the trials in NSCLC summarized here. As noted in Tables 3 and 4, the largest trial randomized 323 patients to one of three different treatment arms [10]. Small but clinically meaningful survival benefits in the NSCLC population were unlikely to be detected by the existing studies, given the inadequate numbers of patients. Also, several of the trials [15, 17, 18] did not reach the accrual goal dictated by the statistical assumption made in the design of the trial. The existing trials, therefore, should not be regarded as conclusive for a lack of benefit to adjuvant therapy and do not provide a definitive answer to the question.

Drug delivery, including both dose and dose intensity, was also inadequate in the majority of studies. The dose of cisplatin in five of the 11 studies was ≤50 mg/m². Table 5 lists the percentages of patients in each of the trials who received no therapy at all (but were randomized to receive it) and the percentage of patients who received all of the planned cycles. In certain studies [20], up to 24% of patients randomized to receive postoperative adjuvant chemotherapy never received it. In those studies that had a defined duration of adjuvant chemotherapy, typically only 50% (range 36%-76%) of patients received the full course of treatment. Given compliance to treatment figures like this, the existing trials hardly seem a fair test of adjuvant chemotherapy in resected NSCLC. In addition, the regimen studied most frequently (CAP) contained agents such as doxorubicin and cyclophosphamide that have inferior single-agent activity compared with a number of newer agents now available. Some of these newer agents, such as vinorelbine [26], paclitaxel [27], and gemcitabine [28], have been shown, in combination with cisplatin, to provide superior survival results or less toxicity in advanced NSCLC. The incorporation of these more active combination regimens into earlier stages of the disease could very well provide enhanced survival. Advances in the outpatient delivery and supportive care of patients utilizing new and more effective antiemetic agents (5HT₃ inhibitors), growth factors (filgrastim), and appropriate hydration schedules may allow more effective delivery of more active drugs than was possible during the era that the majority of these studies were carried out. Also, advances in the field of thoracic surgery have led to shorter recovery times and hospitalization and less morbidity, thereby allowing patients to perhaps be better “chemotherapy candidates” in the postoperative setting.

A comment about the nature of the disease and its treatment is also appropriate. Patients undergoing a major thoracic surgical procedure are often somewhat compromised by that procedure. Although hospitalization times for lobectomy or pneumonectomy range from 4 to 14 days, patients require additional time as outpatients to fully recover. One of the tenets of adjuvant chemotherapy is that the treatment be initiated within a month or so of the curative operation. Since effective therapy in NSCLC is currently considered to be cisplatin-based, patients are exposed to the potential toxicity of these regimens while in a somewhat compromised state. This contrasts sharply to the adjuvant treatments for breast or colon cancer, where the operations are less compromising and the chemotherapy regimens less toxic, thereby allowing more effective delivery. Indeed, we have noted the poor compliance in the 11 trials we have described. Recent advances in supportive care during chemotherapy should greatly enhance compliance. Thus, we believe that the opinion of many physicians that adjuvant chemotherapy is not effective in resected NSCLC is unwarranted.

Current Ongoing Trials of Adjuvant Chemotherapy in NSCLC

The role of adjuvant chemotherapy with or without thoracic radiation therapy (TRT) is continuing to be studied in curatively resected patients with NSCLC (Table 6). A North American Intergroup trial has recently completed accrual. In that trial, approximately 462 patients were enrolled after complete resections for pathological Stage II (T_1-2N₁M₀) and IIIA (T_1-2N₂M₀;T₃N_1-2M₀) disease. Patients were randomly allocated to either TRT alone or concurrent cisplatin/etoposide and TRT. The primary objectives of the study are to determine if chemotherapy plus TRT is superior to TRT alone in improving survival and preventing local recurrences. The trial was designed to detect a 40% increase in median survival for the combination arm with a sensitivity of 0.852. As noted, this trial recently completed accrual and awaits analysis.

The National Cancer Institute of Canada (NCI-C) has initiated a trial in Stage I (T₂N₀) and Stage II (T_1-2N₁) completely resected NSCLC patients (BR10). This trial randomly allocates patients to either observation alone or four cycles of vinorelbine and cisplatin. This trial is also prospectively collecting tissue samples to identify important biologic prognostic factors. Patients entered onto this trial are stratified by nodal status (N₀ versus N₁) and whether a ras mutation is present in their tumor. The primary endpoint is survival. The study is designed to detect an absolute improvement of 10% survival at three years, with an expected 60% survival rate in the observation arm and an exponential survival distribution. In order to have an 80% chance of detecting this difference in overall survival using a log-rank test, with a one-sided 5% alpha level test, 600 patients will need to be accrued over approximately five-years. Both the Southwest Oncology Group (SWOG) and the Eastern Cooperative Oncology Group (ECOG) have elected to participate with the NCI-C in completing this very important trial.

The Montreal Multicenter Trial enrolled completely resected patients from 1984 to 1992, accruing 430 patients. Patients without lymph node invasion were randomized to observation alone versus vindesine/cisplatin, while patients with lymph node involvement received post-operative TRT with or without vindesine/cisplatin. This study has been published in abstract form only and no survival data were given. The investigators also examined predictors of relapse such as lymph node metastases, vascular invasion, chemotherapy dose intensity, and the presence of p53 abnormalities.

The Adjuvant Lung Project Italy (ALPI) trial is a randomized trial for Stages I, II, and IIIA patients who have undergone a curative resection. Patients are randomized to postoperative mitomycin, vindesine, and cisplatin (MVP) versus observation with TRT at the discretion of the treating physician. The study is designed to detect a 20% reduction in mortality following adjuvant MVP with 80% power at the 5% level of significance. It is estimated by the investigators that 1,500 patients will be required, and approximately 550 patients were enrolled onto the study as of May 1997. The European Organization for the Research and Treatment of Cancer (EORTC) is also participating in this study.

The International Adjuvant Lung Cancer Trial (IALT) is a large multicenter randomized trial currently being conducted worldwide. This trial will assess the impact of three or four cycles of cisplatin-based chemotherapy on the survival in curatively resected Stages I, II and IIIA NSCLC. The trial will accrue approximately 3,000 patients. A reduction in mortality of 5%-10% in curatively resected NSCLC is the assumption used in the design of this trial. Given the planned number of patients to be accrued, this trial will have substantially more statistical power than currently published trials of adjuvant therapy in NSCLC.

The British Thoracic Society is conducting The Big Lung Trial (BLT) examining the impact of three cycles of cisplatin-based chemotherapy on the survival of curatively resected NSCLC. Patients may also receive postoperative radiation therapy. In patients undergoing surgery alone, 4,000 will be randomized to receive cisplatin-based chemotherapy or observation. This will be sufficient to detect a 5% improvement in five-year survival (50% to 55%) with at least 90% power. In patients undergoing surgery plus radiotherapy, 2,500 patients will be randomized to receive cisplatin-based chemotherapy or observation. This will be sufficient to detect a 5% improvement in five-year survival (20% to 25%) with at least 90% power. The total number of 6,500 patients to be included in this trial will make it the largest trial to date examining the role of cisplatin-based chemotherapy in adjuvant therapy of NSCLC.

The CALGB has recently activated a trial in Stage I NSCLC including only T₂N₀ patients. The randomization in this study is between adjuvant carboplatin/paclitaxel (four cycles) and observation. The primary objective of the study is to determine a survival benefit of the chemotherapy. The study is designed with an 80% power to detect a 13% increase in the five-year survival and will require 500 patients. It is anticipated that accrual will be completed in approximately three years.

Conclusions

Our analysis suggests that there may be a survival benefit to curatively resected patients who receive adjuvant cisplatin-based chemotherapy. However, the studies reported to date do not yield sufficient data to definitively conclude this. These studies suffer from small numbers of heterogeneous patient populations, suboptimal chemotherapy, inadequate chemotherapy delivery, and very often incomplete accrual. Current ongoing trials have been designed to accrue substantially larger numbers of patients and will hopefully meet their accrual goals. Moreover, they utilize modern cisplatin-based regimens and modern standards of supportive care, thereby providing more reliable data on the effect of adjuvant chemotherapy in curatively resected NSCLC. The recent meta-analysis [9] suggested a 6% improvement in five-year survival in early stage NSCLC from the use of chemotherapy as an adjunct to surgery. Using incidence and percent distribution figures from 1997 [1], approximately 30,000 patients will undergo a curative resection for early stage NSCLC in the United States. A 6% improvement in five-year survival would result in approximately 2,000 lives saved as a result of chemotherapy if this 6% figure is real. Although current reported data are suggestive, it does not definitively tell us that adjuvant chemotherapy should be the standard of care. Given the large number of lives at risk, all eligible patients should be placed on adjuvant chemotherapy trials to help answer this very important question.

References