Annals of Oncology Advance Access published September 5, 2014 1 Adjuvant cisplatin-based chemotherapy in non-small-cell lung cancer: new insights into the effect on failure type via a multistate approach F. Rotolo 1, A. Dunant 1, T. Le Chevalier 2, J.-P. Pignon 1, R. Arriagada 3, on behalf of the IALT Collaborative Group 1 Department of Biostatistics and Epidemiology, Gustave Roussy, Villejuif, France 2 Department of Medical Oncology, Gustave Roussy, Villejuif, France 3 Department of Radiation Oncology, Gustave Roussy, Villejuif, France Corresponding author: Mr Federico Rotolo, Service de Biostatistique et d Epidémiologie, Gustave Roussy, 114 Rue Edouard-Vaillant 94805 Villejuif Cedex (France), Phone: +33 (0)1 42 11 61 28 federico.rotolo@gustaveroussy.fr The Author 2014. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For permissions, please email: journals.permissions@oup.com.
2 Abstract Background. Adjuvant cisplatin-based chemotherapy has become the standard therapy against resected non-small cell lung cancer (NSCLC). Because of variable results on its late effect, we reanalyze the long-term data of the International Adjuvant Lung Cancer Trial (IALT) to describe in details the role of adjuvant chemotherapy. Patients and Methods. In the International Adjuvant Lung Cancer Trial, 1867 patients were randomized between adjuvant cisplatin-based chemotherapy and control, who were followed-up for a median of 7.5 years. Of these, 1687 patients were enrolled from 132 centers accepting to report the times to cancer events. We used event-history methodology to estimate the effects of adjuvant chemotherapy on the risks of local relapse, distant metastasis, and death. Results. Adjuvant chemotherapy was highly effective against local relapses (HR=0.73; 95% CI: 0.60 0.90; P=0.003) and non-brain metastases (HR=0.79; 95% CI: 0.66 0.94; P=0.008) but not against brain metastases (HR=1.1; 95% CI: 0.82 1.4; P=0.61). The effect on non-cancer mortality was non-significant during the first five years (HR=1.1; 95% CI: 0.81 1.5; P=0.29), whereas the risk of non-cancer mortality was subsequently higher with treatment (HR=3.6; 95% CI: 2.2 5.9; P<0.001). This harmful effect, however, potentially concerned only about 2% of the patients at eight years. Conclusion. Adjuvant cisplatin-based chemotherapy reduced the risk of local relapse and of non-brain metastasis, thereby improving survival. This treatment exerted no residual effect on mortality during the first five years, but a higher risk of non-cancer mortality was found thereafter. Detailed long-term follow-up is strongly recommended for all patients in randomized trials evaluating adjuvant treatments in NSCLC.
3 Keywords 1. cisplatin 2. non-small cell lung cancer 3. adjuvant chemotherapy 4. randomized trial 5. multistate model 6. failure pattern Key message Adjuvant cisplatin-based chemotherapy in NSCLC reduces the risk of local relapse and of non-brain metastasis, thereby improving survival. It exerts no residual effect on mortality in the first five years, but is associated to a higher risk of non-cancer mortality thereafter. Detailed long-term follow-up is strongly recommended for all patients in randomized trials evaluating adjuvant treatments in NSCLC
4 Introduction During the last decade, adjuvant cisplatin-based chemotherapy has been established as an effective treatment in resected non-small cell lung cancer (NSCLC) and it has become the standard therapy in this patient population [1 2]. The IALT [3], the first and largest randomized study on this issue, showed a significant overall survival advantage of 4.1% at five years in patients receiving adjuvant chemotherapy compared to surgery alone. The joint LACE analysis (Lung Adjuvant Cisplatin Evaluation [4]) included the IALT and the four other largest randomized trials [5 8]. LACE and, more recently, a worldwide individual patient-based metaanalysis [9] confirmed a survival benefit of about 5% at five years obtained with adjuvant chemotherapy. Even though this improvement was observed at the classic milestone of five years, longterm results should also be explored to ensure that late toxicities do not compromise the initial benefits. Unfortunately, the long-term results of adjuvant studies are rarely known, mainly for logistical reasons. An updated analysis of IALT with a follow-up of 7.5 years [10] confirmed a benefit within the first five years, but showed a 1.5 mortality relative risk thereafter in the treatment arm (95% CI, 1.0 2.1, P=0.04). Is the effect of adjuvant chemotherapy partially counterbalanced in the long term by late deleterious effects or late exacerbation of comorbidities? Two other smaller trials with a median follow-up of 9.3 [11] and 6.2 years [12] however, showed no evidence of a differential effect of chemotherapy over time. One of them used carboplatin [12]. LACE [4] (median follow-up: 5.2 years) could not evaluate the late effects of chemotherapy because long-term data were unavailable for three trials [5 7]. The aim of the present study was to undertake detailed analyses using more sophisticated statistical methodologies, including competing risk and multi-state approaches, to better describe the early and late effects of cisplatin-based adjuvant chemotherapy in resected NSCLC.
5 Materials and methods IALT enrolled 1867 patients with resected NSCLC between February 1995 and January 2001, randomized to cisplatin-based adjuvant chemotherapy or a control arm [3]. They were followed up for a median period of 7.5 years. The participating centers were invited to optionally report, in addition to the time to death, some patient outcomes, including the time to the first local and distant relapse, and to second tumors. Such information can be used to explore the effects of chemotherapy on mortality more thoroughly by separating its antineoplastic effect changes in the risks of recurrence from its residual effect, if any. This potential residual effect could also result from a change in second cancer hazards. We did not consider these events because previous results did not show such an effect [10, 13]. An analysis of the incidence of cancer-related events, based on shorter-term follow-up data, was published earlier [13]. We limited all our analyses to the 1687 patients enrolled from the 132 centers which accepted to report cancer-related events and therefore excluded the 180 patients from the remaining 17 centers (Supplementary Figure S1). Supplementary Table S1 presents a description of the 1687 patients included. We used proportional hazard competing risks and multistate models [14] to estimate the effects of adjuvant chemotherapy on the risks of local relapse, metastasis, and death. We assumed that chemotherapy had a possible effect exclusively during the free-from-any-event period, and not after the occurrence of any cancer-related event. Following previous analyses [13], we assumed different effects of chemotherapy on brain and non-brain metastases. We estimated separately the effect of chemotherapy on mortality before and after five years. Nonetheless, as few cancer-related events occurred after five years, we estimated both the effect of chemotherapy on them and their influence on the overall risk of death for the entire observation period. For full details of the statistical methods, see the online Supplementary Material.
6 Results In total, 1687 patients were followed up for intermediate events. One or more recurrences were reported for 895 patients and none for 792 patients. The database contained information on 401 local relapses and 689 metastases. In total, 1064 patients died during the study, 236 without cancer events, 641 with either a local relapse or metastasis, and 187 patients with both. Figures 1, S2, and S3 show the frequencies of each event sequence. Supplementary Table S2 describes the causes of death by arm, for the whole study and by period. Supplementary Figures S4, S5, and S6 show the evolution along time of the probabilities of being in each clinical state. Local relapses occurred mostly during the first five years: 28% at five years and 32% at ten years. Twenty-six out of the 401 local relapses occurred after a metastasis, but only 15 occurred more than four months later. The risk of local relapse was lower in the chemotherapy arm (HR=0.73; 95% CI: 0.60 0.90; P=0.003). The risk of local relapse was lower after metastasis (HR=0.56; 95% CI: 0.33 0.96; P=0.034; Supplementary Table S3). The risk of death was highly increased after local relapses (HR=26; 95% CI: 20 34; P<0.001). Distant metastases 202 brain and 487 non-brain lesions were even more frequent, with an incidence of 33% at five years and of 38% at ten years. Overall, 63 metastases were observed after local relapses, but we only 32 metastases occurred more than four months later. Chemotherapy was highly effective in reducing non-brain metastases (HR=0.79; 95% CI: 0.66 0.94; P=0.008) but not brain metastases (HR=1.1; 95% CI: 0.82 1.4; P=0.61). The risk of metastasis was significantly higher after local relapse (HR=1.7; 95% CI: 1.2 2.5; P=0.003), estimated without distinguishing brain and non-brain metastases due to the small number of events. The risk of death was highly increased after brain metastasis (HR=38; 95% CI: 29 50; P<0.001) and non-brain metastasis (HR=40; 95% CI: 31 51; P<0.001).
7 We estimated that 53% of the controls developed a local relapse or a non-brain metastasis at eight years. Consequently, more than half of the patients could potentially benefit from the chemotherapy. Once this benefit was taken into account, its residual effect (i.e. on non-cancer mortality) was non-significant during the first five years (HR=1.1; 95% CI: 0.81 1.5; P=0.29), meaning that all the activity of chemotherapy consisted in the prevention of recurrences (excepted brain metastases). In contrast, after five years, the risk of non-cancer death was higher in the treatment arm (HR=3.6; 95% CI: 2.2 5.9; P<0.001), even though the antineoplastic benefit was still present. Of note, only 36% of the controls were alive and free of cancer at five years and only 6% of them (2% of all the controls) died of other causes during the next three years. Figure 2 and Supplementary Table S3 summarize the main results and Supplementary Table S4 presents the effects of the adjustment variables. As expected, the risks of all events, including death, increased with advanced tumor and nodal stage. The risk of death was higher for advanced stages irrespective of whether cancer events had occurred or not (not shown). Patients who had undergone a lobectomy or a segmentectomy, as compared to pneumonectomy, had a higher risk of local relapse (HR=1.3; 95% CI: 1.1 1.5; P=0.003), brain metastasis (HR=1.4; 95% CI: 0.98 1.9; P=0.065), and non-brain metastasis (HR=1.2; 95% CI: 0.97 1.5; P=0.088). A lower risk of death in lobectomy/segmentectomy patients (HR=0.87; 95% CI: 0.78 0.96; P=0.001) was observed, after taking into account the effect of cancerrelated events. An early toxic effect of chemotherapy during the first 12 months for non-cancer deaths (HR=1.7; 95% CI: 1.0 2.8; P=0.044) was observed, consistent with previous findings during the first six months [4]. Moreover, we compared the incidence of non-cancer deaths among patient free from cancer at five years in the treatment arm, between patients with planned total cisplatin total doses of 300 mg/m 2, 320 mg/m 2, 360 mg/m 2, and 400 mg/m 2. The non-cancer mortalities were very similar (log-rank test P=0.57) with an 8-year incidence of 13%, 11%, 13%, and 6%, respectively. Likewise, no link was found between the cisplatin-associated regimen
8 and the late increase in mortality: among 5-year survivors, the 8-year survival was 0.86 (etoposide) vs. 0.87 (vinca alkaloids), with a non-significant difference between their subsequent survival (log-rank test P=0.73). Discussion The benefit yielded by cisplatin-based adjuvant chemotherapy in the IALT was a reduction in the risk of both local relapses and non-brain metastases. It had no effect on the risk of brain metastases, probably due to the blood-brain barrier [15]. Besides this antineoplastic activity, cisplatin-based adjuvant chemotherapy had no additional effect on mortality during the first five years. In contrast, although cancer events declined after the first five years, the risk of non-cancer deaths increased in the treatment arm compared to the control arm. This long-term increased risk for treated patients was relatively high (three-fold), but only 2% of the controls died of non-cancer causes between 5 and 8 years whereas the benefit (a 25% reduction in the risk) concerned cancer events which affected more than 50% of the controls at 8 years. As only 5% of cancer events occurred after five years, we assumed that adjuvant chemotherapy exerted a similar effect on them throughout the study. A complementary analysis with separate estimated effects on the two periods did not show any significant difference in the effect of chemotherapy on cancer events over time (data not shown). The risk of each type of relapse changed significantly after the occurrence of other cancer-related events. The risk of metastases increased by 90% after a local relapse. This result was probably due at least to two different phenomena. Firstly, after a local relapse, the physician prescribed additional examinations to detect metastases, which were therefore diagnosed earlier and more frequently than in patients without a local relapse. Secondly, a local relapse could itself give rise to new metastases unrelated to the primary cancer. Conversely, the risk of local relapse decreased by 44% after a metastasis. A sensible
9 explanation for this risk reduction is that metastases are so life threatening that physicians focused on immediate patient survival rather than local relapse detection. An increased incidence of second malignancies has been observed in testicular cancer patients treated with cisplatin [16 19], but not in IALT [10]. Therefore, we decided to exclude them from the current analyses. As expected, the risk of death was strongly increased after a cancer-related event. The huge hazard ratios (26, 38, and 40) could have been overestimated because the exact time to death is known, whereas cancer events are diagnosed and recorded later than their onset. Thus, shorter relapse-to-death gaps could lead to an overestimation of the risk of death after relapse. Finally, we observed a lower risk of all recurrences in patients who had undergone a pneumonectomy than those submitted to lobectomy or segmentectomy, but a higher risk of death once the effects on the relapses were taken into account. This suggests that excision of the whole lung reduces the risk of relapse but increases mortality due to other causes, which is consistent with clinical knowledge [20 21]. The present analysis is the first to employ cutting-edge statistical methods to study the role of adjuvant cisplatin-based chemotherapy in resected NSCLC. Its main results are in accordance with those published earlier [10]. However, the use of multistate models allowed distinguishing the effects of cisplatin, neatly separating its antineoplastic role from the late increase in non-cancer mortality. In addition, this methodology enabled us to estimate the impact of cancer events on the risk of subsequent events including death. Our major finding is that adjuvant chemotherapy is effective against local relapses and non-brain metastases but that it leads to increased non-cancer mortality after five years. It is well established that patients with resected NSCLC experience excess long-term mortality compared to the general population, probably because of co-morbidities [22]. Nevertheless this would not explain differential long-term non-cancer mortality between treated and control NSCLC patients. In our study, this late over-mortality could be an artifact due to the low number of patients in some transitions in the multistate model. However, the results are
10 consistent with previous analyses with simpler methods. Regarding the potential role of postoperative radiotherapy, it was a center choice and then stratified in the randomizing procedure. In addition, we did not find a late deleterious effect related to radiotherapy [10]. Our findings concur with the increased risk of potentially lethal late effects of platinum-based chemotherapy evidenced in large databases on testicular cancer [16 18, 23 24]. A potential toxic effect of cisplatin-based chemotherapy on non-cancer deaths was recently evaluated by a literature-based meta-analysis including 6430 NSCLC patients from 16 randomized trials including IALT [25]. Over the whole follow-up period, there was a significant increase (30%) in non-cancer mortality in the cisplatin arm but the time of occurrence of the deleterious effect was not studied. When the IALT is excluded, the increased relative risk remains borderline significant (RR=1.2; 95% CI: 0.99 1.48, P=0.06) over a median follow-up of 5.4 to 9.6 years for the three larger studies. Even if this late over-mortality only concerns a small percentage of the patients, it will be higher with future potent antineoplastic adjuvant, unless this improvement also includes a decrease in late toxic effects. The optimal situation would be to find tumor markers able to select chemosensitive tumors, and thus better target the subpopulations that are likely to benefit, but until now this research has remained inconclusive [26 27]. The most plausible explanation for the late over-mortality is a long-term toxic effect of cisplatin suggested by published results in NSCLC and other cancers. Nevertheless, the current results are based on a small number of non-cancer deaths after five years and longterm follow-up in this category of randomized trials is scarce nowadays. Longer and detailed follow-up should therefore be strongly recommended in the ongoing randomized phase III trials evaluating adjuvant NSCLC treatments.
11 Acknowledgements The authors are grateful to Lorna Saint Ange for editing. They thank the Steering Committee and all the IALT Collaborators for providing the IALT data set (Appendix 1). Funding No specific financial support Disclosure The authors have declared no conflict of interest
12 References [1] Pisters KMW, Evans WK, Azzoli CG et al. CCO and the ASCO guideline for adjuvant chemotherapy and adjuvant radiation therapy for stages I-IIIA resectable non-small cell lung cancer. J Clin Oncol 2007; 25: 5506 18. [2] National Comprehensive Cancer Network. NCCN guidelines version 1.2013, Nonsmall cell lung cancer NCCN guidelines for patients. http://www.nccn.org/patients/guidelines/nscl/index.html. Accessed April, 2014. [3] The International Adjuvant Lung Cancer Trial Collaborative Group. Cisplatin-based adjuvant chemotherapy in patients with completely resected non-small-cell lung cancer. N Engl J Med 2004; 350:351 60. [4] Pignon JP, Tribodet H, Scagliotti GV et al. Lung Adjuvant Cisplatin Evaluation: A Pooled Analysis by LACE Collaborative Group. J Clin Oncol 2008; 26:3552 8. [5] Douillard JY, Rosell R, De Lena M et al. Adjuvant vinorelbine plus cisplatin versus observation in patients with completely resected stage IB IIIA non-small-cell lung cancer (Adjuvant Navelbine International Trialist Association [ANITA]): a randomised controlled trial. Lancet Oncol 2006; 7: 719 27. [6] Scagliotti GV, Fossati R, Torri V et al. Randomized study of adjuvant chemotherapy for completely resected stage I, II or IIIa non-small cell lung cancer. J Natl Cancer Inst 2003; 95: 1453 61.
13 [7] Waller D, Peake MD, Stephens RJ et al. Chemotherapy for patients with non-small cell lung cancer: the surgical setting of the Big Lung Trial. Eur J Cardiothorac Surg 2004; 26: 173 82. [8] Winton T, Livingston R, Johnson D et al. Vinorelbine plus Cisplatin vs. observation in resected non-small cell lung cancer. N Engl J Med 2005; 352: 2589 97. [9] NSCLC Meta-analyses Collaborative Group. Adjuvant chemotherapy, with or without postoperative radiotherapy, in operable non-small cell lung cancer: two meta-analyses of individual patient data. Lancet 2010; 375: 1267 77. [10] Arriagada R, Dunant A, Pignon JP et al. Long-term results of the international adjuvant lung trial evaluating adjuvant Cisplatin-based adjuvant chemotherapy in resected lung cancer. J Clin Oncol 2010; 28:35 42. [11] Butts CA, Ding K, Seymour L et al. Randomized Phase III Trial of Vinorelbine Plus Cisplatin Compared With Observation in Completely Resected Stage IB and II Non Small-Cell Lung Cancer: Updated Survival Analysis of JBR-10. J Clin Oncol 2010; 28: 29 34. [12] Strauss GM, Herndon JE, Maddaus MA et al. Adjuvant paclitaxel plus carboplatin compared with observation in stage IB non-small-cell lung cancer: CALGB 9633 with the Cancer and Leukemia Group B, Radiation Therapy Oncology Group, and North Central Cancer Treatment Group Study Groups. J Clin Oncol 2008; 26: 5043 51. [13] Dunant A, Pignon JP, Le Chevalier T. Adjuvant Chemotherapy of Non-small Cell Lung Cancer: Contribution of the International Adjuvant Lung Trial. Clin Cancer Res 2005; 11: 5017s 21s.
14 [14] Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: competing risks and multistate models. Stat Med 2007; 26: 2389 430. [15] Hansen HH. Should initial treatment of small cell carcinoma include systemic chemotherapy and brain irradiation? Cancer Chemother Rep 1973; 4:239 41. [16] van den Belt-Dusebout AW, de Wit R, Gietema JA et al. Treatment-Specific Risks of Second Malignancies and Cardiovascular Disease in 5-Year Survivors of Testicular Cancer. J Clin Oncol 2007; 25: 4370 8. [17] Travis LB, Fosså SD, Schonfeld SJ et al. Second cancers among 40,576 testicular cancer patients: focus on long-term survivors. J Natl Cancer Inst 2005; 97: 1354 65. [18] Haugnes HS, Bosl GJ, Boer H et al. Long-term and late effects of germ cell testicular cancer treatment and implications for follow-up. J Clin Oncol 2012; 30: 3752 63. [19] Travis LB, Andersson M, Gospodarowicz M et al. Treatment-associated leukemia following testicular cancer. J Natl Cancer Inst 2000; 92: 1165 71. [20] Okada M, Yamagishi H, Satake S et al. Survival related to lymph node involvement in lung cancer after sleeve lobectomy compared with pneumonectomy. J Thorac Cardiov Sur 2000; 119: 814 9. [21] Deslauriers J, Grégoire J, Jacques LF et al. Sleeve lobectomy versus pneumonectomy for lung cancer: a comparative analysis of survival and sites or recurrences. Ann Thorac Surg 2004; 77: 1152 6.
15 [22] Janssen-Heijnen ML, van Steenbergen LN, Steyerberg E et al. Long-term excess mortality for survivors of non-small cell lung cancer in the Netherlands. J Thorac Oncol 2012; 7:496 502. [23] Meinardi MT, Gietema JA, van der Graaf WT et al. Cardiovascular morbidity in long-term survivors of metastatic testicular cancer. J Clin Oncol 2000;18: 1725 32. [24] Fosså SD, Gilbert E, Dores GM et al: Noncancer causes of death in survivors of testicular cancer. J Natl Cancer Inst 2007; 99:533 44. [25] Petrelli F, Barni S. Non-cancer-related mortality after cisplatin-based adjuvant chemotherapy for non-small cell lung cancer a study-level meta-analysis of 16 randomized trials. Med Oncol 2013; 30: 641. [26] Olaussen KA, Dunant A, Fouret P et al. DNA repair by ERCC1 in non small-cell lung cancer and cisplatin-based adjuvant chemotherapy. N Engl J Med 2006; 355(10), 983 91. [27] Friboulet L, Olaussen KA, Pignon JP et al. ERCC1 Isoform Expression and DNA Repair in Non Small-Cell Lung Cancer. N Engl J Med 2013; 368(12), 1101 10.
16 Figure legends Figure 1. Number of events and event sequences in the IALT study both arms. For example: a total of 202 bdm were observed, 177 as first event, 12 concomitantly with LR, and 13 after LR; among the 177 patients with bdm only, 7 had LR later, 155 died, and 15 were censored; among the 32 patients with both bdm and LR, 31 died and 1 was censored. DM: distant metastasis; LR: local relapse. The numbers in the grey boxes are the number of patients whose follow-up data were censored (c=). Note that only the time to the first metastasis, either brain or non-brain, was available for each patient. Events on the arrows to intersections of event sets (for instance those 12 patients from randomization to LR+bDM) denote simultaneous events; other events occurred sequentially, like those 7 patients having LR after bdm. Figure 2. Summary of results (hazard ratios) regarding the chemotherapy (CT) effect as compared to control (Ctrl) and the effect of events on each other, from the three multistate models: (1) the model for death with intermediate recurrences (solid arrows), which also estimated the residual (= remaining after taking into account the effect of chemotherapy on recurrences) effect of chemotherapy on death, before and after five years. (2) The model for local relapse (LR), with non-simultaneous intermediate distant metastases (DM, dotted arrows). (3) The model for brain and non-brain metastasis (bdm, nbdm), with nonsimultaneous intermediate local relapse (dashed arrows). Grey-colored HRs are not significantly different from 1 at a 95% level. Black HRs are significantly lower/higher than 1. Significance codes: *** = <0.001, ** = 0.001 to 0.01, * = 0.01 to 0.05
17
18