This review investigated the effect on survival from resectable non-small-cell lung cancer (NSCLC) of induction or adjuvant chemotherapy compared with no chemotherapy. The authors found that adjuvant chemotherapy improved survival in patients with stage I and II NSCLC. Despite some shortcomings in the review methodology, the results are likely to be reliable.

To determine the efficacy of induction and adjuvant chemotherapy in patients with resectable non-small-cell lung cancer (NSCLC).

MEDLINE, HealthSTAR and the National Cancer Institute's databases were searched from 1965 to June 2004 for relevant papers published in English or French. Further searches of bibliographies, textbooks and reviews were performed. Relevant studies of which the investigators were aware were also included. Although abstracts were not eligible for inclusion, two trials presented at the 2004 ASCO meeting were included as additional information was available from the conference presentations.

Study designs of evaluations included in the review

Phase III prospective, randomised clinical trials were eligible for inclusion.

Specific interventions included in the review

Studies of chemotherapy given before or after surgery were eligible for inclusion.

For induction chemotherapy, the included interventions were combinations of cisplatin, etoposide, mitomycin C, ifosfamide, cyclophosphamide and vindesin. Radiotherapy was offered to patients with incomplete resection or late-stage disease. The control interventions were radiotherapy or no additional treatment.

For adjuvant chemotherapy, the included interventions were combinations of adriamycin, cyclophosphamide, cisplatin, vindesin, vinblastin, vincristin, vinorelbin and tegafur plus uracil. Radiotherapy was offered to patients with incomplete resection or late-stage disease. The control interventions were radiotherapy, bacillus Calmette-Guerin plus levamisole, or no additional treatment.

Participants included in the review

Studies of patients with resectable NSCLC were eligible for inclusion. The included patients had stage I, II or III NSCLC. No further details of the participants were given.

Outcomes assessed in the review

No inclusion criteria for the outcomes were stated. The outcome of interest in the review was survival.

How were decisions on the relevance of primary studies made?

The authors did not state how the papers were selected for the review, or how many reviewers performed the selection.

Six doctors and one statistician assessed validity, with any disagreements resolved by consensus between at least five of them. Two scales were used to assess validity: the European Lung Cancer Working Party (ELCWP) scale (based on protocol design and study analysis report) and the Chalmers scale (based on internal and external validity).

The authors did not state how many reviewers performed the data extraction. For each review, the hazard ratio (HR) was extracted or estimated from: the HR point estimate, the log rank statistic or its p-value, the difference between observed and expected events (O-E) statistic or its variance;

the total number of events and its log rank statistic or its p-value; or

published survival curves.

How were the studies combined?

The studies were combined using Peto's fixed-effect method for meta-analysis. Where there was statistically significant heterogeneity (p<0.10), a random-effects model was used.

How were differences between studies investigated?

Tests of heterogeneity were performed. Subgroup analyses were performed by stage of disease (stage I and II versus III), type of chemotherapy (cisplatin and tegafur plus uracil) and radiotherapy use (yes versus no).

A total of 25 trials (8,234 participants) were included in the review: 6 trials (590 participants) investigated induction chemotherapy and 19 trials (7,644 participants) investigated adjuvant chemotherapy.

The mean ELCWP and Chalmers scores for validity were 67.8% (range: 48.9 to 92.5) and 54.6% (range: 37.5 to 86.0), respectively. The validity scores from both scales were similar between trials of different sizes, those investigating induction versus adjuvant chemotherapy or cisplatin versus other chemotherapy, those including patients with early- or late-stage disease, and those reporting a statistically significant effect of the intervention or not.

Induction chemotherapy was significantly associated with improved survival (HR 0.66, 95% confidence interval, CI: 0.48, 0.93) when all disease stages were pooled; there was a suggestion of statistical heterogeneity (p=0.07). Among patients with stage III NSCLC, induction chemotherapy was associated with a non significant improvement in survival (HR 0.65, 95% CI: 0.41, 1.04); there was evidence of statistical heterogeneity (p=0.02).

Adjuvant chemotherapy was significantly associated with improved survival (HR 0.84, 95% CI: 0.78, 0.89); there was no evidence of statistical heterogeneity (p=0.24). The addition of two trials only reported at a conference did not alter this effect. Similar effects were found across all strata in the subgroup analyses.

Adjuvant chemotherapy improved survival in patients with stage I and II NSCLC. More data are needed to confirm such a role for induction chemotherapy. Further trials should consider stage III and earlier stage disease separately.

The search strategy was not fully described, and might not have identified all relevant studies, owing to the omission of those published in languages other than English or French. In addition, there was no attempt to search for unpublished reports. Therefore, the synthesised results may be biased by publication or language bias. Since inclusion criteria for the outcomes were not stated, and there were no details of how the relevance of included studies was assessed, subjective decisions might have been made during the search. A validity assessment was performed. The evidence that study quality was not related to whether the study reported a beneficial result of the intervention or not suggests that the results are likely to be reliable.

There were insufficient details of the participants in the individual trials to assess the external generalisability of the studies. The meta-analytic process appeared adequate and, although no statistical attempt was made to explain the presence of statistical heterogeneity between the studies, possible sources of heterogeneity were discussed.

It was unclear why the authors concluded that adjuvant chemotherapy improved survival but that more evidence was required for induction chemotherapy, since the effect in the latter group was stronger than that in the former group, based on the magnitude of the HR. It was noted that the number of participants in the induction chemotherapy group was substantially smaller than in the adjuvant group. The magnitude of the effect of adjuvant chemotherapy on survival was almost identical for stage I and II versus stage III patients; it appeared that the authors based their conclusions on whether the results reached conventional levels of statistical significance or not, although this was not explicitly stated.

Practice: The authors recommended the use of a platinum-based regimen in addition to surgery outside the context of clinical trials.

Research: The authors stated that future trials should consider patients with stage I and II NSCLC separately from those with stage III disease, and recommended further research on stage III disease. Further studies are required to confirm the results of induction chemotherapy. Outstanding research questions were whether there was a difference in effect of induction versus adjuvant chemotherapy in stage IB and II NSCLC, the efficacy of chemotherapy in stage IA disease, and the comparison of differing chemotherapy regimens.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.